Strange dilating behaviour

[Nevsden]

Using a positive offset, the algorithm computes a dilated offset. Instead of using a rounding approach to dilate, it instead uses a kind of star-shaped approach. Is this behaviour desired?

Note, that the original two dimensional line is the fifth counting from the innermost line. The four inner lines are computed using the shrinking offset. From the original line, all other lines are computed using the offset algorithm.