Hi PaQ,

I cant give you an exact reason. I just trial & error. But I got the result below. Select all curves use window selection, then pick network. I think Michael or manz will give you a better explanation. Me is one of the moi user that still confuse how to use this feature correctly.

@ Anis : you have only 5 curves here!
Original has 6 curves! A little at the nose exists ;)

Yup !!!

Join, then trim.
See below...

Yes but this is defined as number 1 on the original drawing ;)
I believe that PaQ want to keep it for test the Network function ;) ( ? )

@Pilou...

Upsss....
OK, I see now. So lets wait an expert explain this. I wish some body can give us how network will work or not in a situation. Better if include with illustration image.

Sorry guys...

Hi PaQ - the picking order doesn't have any effect on Network, it depends on looking at all the intersections between the curves to sort things out.

It doesn't look like there is a problem in your layout, so this might be a bug, I'll take a look.

One quick other note though - it can cause problems to have a surface edge that comes to a tangent with its neighboring edges like you have at the small tip there. The surface-normal direction of a surface is usually defined as a perpendicular to 2 directions of the surface at any point. When you have that kind of a situation, the 2 surface directions are colinear, it makes the normal not well defined in that spot.

For operations that depend on the normal such as offsets that will cause a problem.

- Michael

Hi PaQ,

Here is my solution,

Hi PaQ - the bug in this case has to do with the small size of the curve at the tip.

It is less than 1/100 the size of the bounding box around all the curves, which is a tolerance that is used to determine if all the curves in a direction are coming to a shared single "pole" type point.

So it is getting confused and thinking that small piece is part of a "pole" arrangement with the other 3 long length-wise pieces.

I have now fixed this up for the next beta so that it will also take curve length into account when computing the pole tolerance, that makes this case then work fine.

For the moment you'll need to leave out that small one at the tip to make it work.

- Michael

Alright, I'm happy to see it was just a bug (and that it was already fixed :))

Thanks everyone for the different workarounds too !!!