ExtendSrf has a fundamental flaw when used to extend a trimmed, multi-span surface.
Extending a trimmed, multi-span surface does not follow the untrimmed base surface across spans. Instead it appears that the trimmed surface is first shrunk,. Then the new base surface is extended and finally trimmed. Note: This thread is intended to be a report of the flaw and hopefully get it on the list to be fixed or to provide an option for the expected result, not how to work around it.
Example: ExtendSrfTrimmedFlawDC01.3dm (196.8 KB)
Untrimmed original surface:
Trimmed, but not shrunk, surface. The untrimmed base surface is the same as the original surface:
ExtendSrf trimmed edge. Result does not follow the untrimmed base surface as expected:
Expected result. Extended surface follows the untrimmed base surface:
My guess is the trimmed surface is shrunk before extending to reduce the number of knots/spans. The shrunk base surface is then extended, the new trim curves in the extended position are determined, and the extended base surface is trimmed. I remember seeing something about reducing the number of knots in extended, trimmed fillet surfaces.