The idea underlying our approach is simple: if a user draws an approximately straight line, his intention is to draw a truly straight line; if he draws an approximately parallel set of approximately straight and approximately equally long lines, he wants a set of truly straight, truly parallel and truly equally long lines.