Hi Martin, so I think I've finally gotten to the bottom of what was going wrong with the Inset case that you posted in this thread - it had to do with a bug in the "test for coincidence between closed curves" mechanism in the geometry library.

I've got it fixed for the next v3 beta and it looks like this case that was behaving with sporadic errors should work properly after that.

Thanks for posting the example file.

- Michael