Re: help with coverity spotted problem in region.c::CreatePolyPolygonRgn

[James McKenzie]

Marcus Meissner wrote:
> Hi,
>
> Coverity says something about array indices in CreatePolyPolygonRgn
> and I see some code that looks bad:
> if (((nbpolygons == 1) && ((*Count == 4) ||
>    ((*Count == 5) && (Pts[4].x == Pts[0].x) && (Pts[4].y == Pts[0].y))) &&
>     (((Pts[0].y == Pts[1].y) &&
>       (Pts[1].x == Pts[2].x) &&
>       (Pts[2].y == Pts[3].y) &&
>       (Pts[3].x == Pts[0].x)) ||
>      ((Pts[0].x == Pts[1].x) &&
>       (Pts[1].y == Pts[2].y) &&
>       (Pts[2].x == Pts[3].x) &&
>       (Pts[3].y == Pts[0].y)))))
>
> The braces seems to be mismatched, so that even for
> *Count==4 it dereferences Pts[4].
>
> But I cannot really make sense of the Count==5 version either, 
> it seems broken too. Pts[4] is compared to Pts[0] points, but then
> again the 4 point polygon is compared.
>
> Anyone? git blame says its from Alexandre in 1998 ...

Marcus:

At first look, there appears to be an extra set of paraenthesis in
this.  Remove the first left and last right.

James McKenzie