Hello, > > >There's no need for different meanings at different times; we could just > > >allow labels in the arms of COND_EXPR at all times, and say that the goto > > >is implied. > > No. > > dest = test ? &&L1 : &&L2 but this should not have a void type, right? Zdenek