[Freeciv-Dev] Re: common/city.h

[Raimar Falke <hawk@xxxxxxxxxxxxxxxxxxxxxxx>]

On Sun, Oct 14, 2001 at 01:01:56AM +0100, Gaute B Strokkenes wrote:
> On Mon, 24 Sep 2001, stoecker@xxxxxxxx wrote:
> > On Mon, 24 Sep 2001, Raimar Falke wrote:
> > 
> >> > > You won't catch the wrong tests in the ai code. Either you
> >> > > audit the code or add checks to ensure that there are no
> >> > > uninitialsed type fields or other nasty things.
> >> > 
> >> > If I understand it correctly, this was left in for security
> >> > reasons, when these enums where introduced. So "theoretically"
> >> > there should be no problems. Practically there maybe some, but we
> >> > never find them without removing the =0. At least there should be
> >> > none, which can be found by easy code reviews.
> >> 
> >> This at ./ai/aicity.c:416: "else if (!bestchoice.type &&
> > 
> > Line 411 according to web interface.
> > 
> >> is_wonder(bestchoice.choice)" is one of the problems which is
> >> exists in the code. At least I can't tell if "!bestchoice.type"
> >> should be "bestchoice.type==CT_NONE" or
> >> "bestchoice.type==CT_BUILDING". So it is not as easy as removing
> >> the "=0".
> > 
> > Compared to line 255 I would say CT_BUILDING.
> > 
> > But I see, that there are far too much positions like this left. I
> > thought the progress was gone farther.
> 
> What about using
> 
> #define CT_NONE 0
> 
> here?  The principal benefit of using an enum over defines is that you
> can see the names in a debugger.  That falls down when there are
> multiple names for the same value...

So you band-aid the bug instead of fixing it? Bad. Very bad.

        Raimar

-- 
 email: rf13@xxxxxxxxxxxxxxxxx
 "Your mail could not be delivered to the following Address:
  VTCMC.VTLPR@xxxxxxxxxxxxx        ** Unassigned error message **"