Reasons for certain inaccuracies in the majority of existing computer programs that numerically calculate transonic full potential flows are clearly exposed. Problems of post-shock oscillations, non-unique shock strengths and their locations, diverging solutions of choked nozzle flows and flows about sharp leading edges are shown to be related to widely accepted oversimplifications in the treatment of boundary conditions. Forms of numerical dissipation that are presently widely used are shown to be not fully understood and an alternative dissipation generation concept is proposed.
