> I hope you verified that. > > I passed the report on to the authors since I cannot see anything > oviously wrong. I verified those bad results (cos == sin) on powerpc. If that has broken at all recently, I can't see how.