Hi wim, re: sweep problem - it's another example of the "teeny tiny object size" problem, here the width of your sweep profile is 0.001 units across, if you scale this up by even 10 times in size it should then work ok as in the attached version.

Basically some operations do additional refinement of their calculations until their result is within 0.001 units of the ideal result. But when your entire shape is also only 0.001 units across, that same 0.001 accuracy target just doesn't provide enough accuracy for things to give good results.

Later on I'll see if I can make these cases that you ran into today work with relative tolerance. But it can be a tricky thing to do very well, especially when it's not the entire bounding box size that's small but only one single piece.

At any rate, it's again a case of if you don't make things with much smaller than say 0.01 you'll get more reliable operation for this.

- Michael