Hi Mark, thanks for posting the model as well, that really helps to see what is happening.

Burr definitely has the right answer there - during the trim the intersection points were just a teeny tiny bit off so the circle piece you got was not actually a 180 degree piece, it is just slightly over that, so that's where the outline is not closed.

Just in case you have any problems following what Burr said, I have attached an updated file here which should now cap for you.


I can repeat the problem with Trim over here, so I'll take a look at what I can do to make that work better.

- Michael

Thanks, guys. That kept me going.

Hi Mark, for the past several days I've been looking into this problem you had with just a rectangle and a tangent circle trimming each other, and I've got several fixes in this area for the next beta release.

To start with, there was a bug in Join introduced in v2 where it would not close off the very first and last segment in the loop in certain conditions (when other pieces joined at a tight tolerance and the first/last segment were at a looser tolerance level but still within the maximum join distance).

I think that was the main problem here, and that one is fixed up for the next beta.


I also added some other stuff in this area as well - lines, circles, and arcs will now be handled using a special analytic curve class which is able to handle intersections between lines and circles with a special case handler that has greater accuracy.


I also added a different special case handler to the Circle/Arc Tangent command for the case of putting in a tangent circle inside of a rectangle, to calculate that case directly (with higher accuracy) rather than going through the iterative solver mechanism.


So I think these things will help to avoid this problem you ran into here, at least as long as there aren't any new bugs introduced with them that is.

- Michael