[freeciv-ai] Re: ai move cleanup #4

[Gregory Berkolaiko <Gregory.Berkolaiko@xxxxxxxxxxxx>]

On Thu, 8 Aug 2002, Per I. Mathisen wrote:

> On Thu, 8 Aug 2002, Gregory Berkolaiko wrote:
> > Issue 1:
> >
> > In ai_unit_bodyguard_move:
> > +if (bodyguard->moves_left < map_move_cost(bodyguard, x, y)) {
> > +  /* should generally should not happen */
> > +  freelog(LOG_VERBOSE, "%s: %s left its %s bodyguard behind at (%d,%d)!",
> > +          pplayer->name, unit_type(punit)->name,
> > +          unit_type(bodyguard)->name, x, y);
> > +  return;
> > +}
> >
> > If bodyguard->moves_left > 0, the bodyguard can still attempt to move!
> 
> Right.
> 
> (Interestingly, if I read the code right, a unit with move == 0 can still
> attack. Is this correct or a misreading or a bug? Seems the client stops
> it from happening, but a modified client should be able to do it.)

Where did you find it?  That'd be very strange.

> > Issue 3:
> >
> > In do_unit_goto, why do you remove
> >
> > -    /* Don't attack more than once per goto */
> > -    if (penemy && !pplayer->ai.control) { /* Should I cancel for ai's too? 
> > */
> > -   punit->activity = ACTIVITY_IDLE;
> > -   send_unit_info(NULL, punit);
> > -   return GR_FOUGHT;
> > -    }
> > -
> 
> This check is done in ai_unit_move. The function returns FALSE, and
> do_unit_goto returns GR_FAILED. Just as it was before, except then it was
> handle_unit_move_request that did it. The AI code doesn't check for
> GR_FOUGHT and the above code would (in theory) never be called.

Just a moment.  The above code is for the humans!  It should stay IMO.

> > The rest seems to be pretty good work.Also seems to fix the bug when an
> > attacker wouldn't occupy an empty city because it's bodyguard wouldn't
> > budge.I didn't test it though.  Savegames would differ at later stage, I
> > assume?

What happened to the spaces after full stops in the passage above??

G.