For example: a skewed bridge. It is very easy to create a shape matching the bridge abutments or construct the underpass corridor and use it for the clipping boundary. The model is clipped perfectly. However, the geometry generated by the corridor should also be clipped.
Example 2: Intersections - if using clipping references for intersections then again, the model is clipped perfectly. But if there is a shoulder line (or other lines) generated by the mainline corridor these should also be clipped.
Example 1 is the more serious case because such skewed clip areas require fairly draconian workarounds. For such skewed clips, using a clipping reference is really the only solution. Then the workaround to manage clipping of the geometry invariably involves copying the geometry. While copy then clip is not really hard, it does raise potential problems related to keeping the copies in sync as changes occur.
Example 2, while important, can be worked around more easily. And in fact, In my opinion at least, the workarounds (parametric constraints is my preference) are actually more efficient in the long run because the corridor processing for these workarounds is less intense than a clip.
My sentiments exactly.
Charles (Chuck) Rheault CADD Manager
MDOT State Highway Administration