Thanks guys. Personally I third your opinions, but wanted to bring it up. cheers Mattias
On Wed, Sep 2, 2015 at 9:56 AM, Sebastian Messerschmidt <[email protected]> wrote: > Hi Matthias, > > I second Robert's opinion. Passing a zero-length vector simply causes in > invalid results to be calculated. > My vote would be towards using an assert, so the user is noticed in debug > builds about the potentially wrong results. > Thats what some other libraries do to hint invalid parameters in > mathematical functions at least in checked/debug builds.. > > > Cheers > Sebastian >> >> Hi >> >> Our team chased down a bug where we passed an invalid rotation vector to: >> >> void osg::Quat::makeRotate(const osg::Vec3d& from, const osg::Vec3d& to) >> >> we passed an all zeroes Vec3d in the from parameter. This is then used >> to normalize the vector, which causes INF values and lots of bad >> things further down the line in our code. I know the bug is in our >> code but I also see that there are divide-by-zero checks made in other >> variations of makeRotate(...). The makeRotate_original even had a >> warning print-out. >> >> I guess it's to old performance vs user-friendliness problem, and >> that's why I wanted to bring it up here before submitting anything. >> >> What do you think? >> >> /Mattias >> _______________________________________________ >> osg-users mailing list >> [email protected] >> http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org > > > _______________________________________________ > osg-users mailing list > [email protected] > http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org _______________________________________________ osg-users mailing list [email protected] http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
