Re: avoiding cppunit zerodiv correction

On Fri, May 30, 2014 at 10:01:49AM +0200, Matteo Casalin wrote:

> > I wonder if we should not rather do something like:

> > double dRatioPreV;
> > if( nHeight < 0 )
> > {
> >   nHeight = 0;
> >   dRatioPreV = 1; // or rather zero? or +infinity? Do we want to
> >                   // "force" a branch of the next if/else?
> > }
> > else
> >   dRatioPreV = ((double) nWidth ) / nHeight;

> > Anybody has a good idea, preferably based on an understanding of
> > what this function does?

> I don't have any comment about your proposal, but just would like to
> add that unless we know for sure that nHeight!=0 at that point, in
> both codes the check should be

> if( nHeight <= 0 )

> in order to solve the division by 0 completely.

Ah yes, good point.

-- 
Lionel