Bug w/curveboolean?

I end up with 2 extra lines when I do this curve boolean, even though I’ve snapped to the quadrants. Doesn’t always happen though. Sometimes I end up with 1 extra, sometimes none.

Here’s another video where I do the boolean & undo it multiple times. You can see that the resulting number of curves is different, seemingly at random. Scratch that, it alternates between 4 & 6 curves when exploded, so I guess not random.

Without a .3dm file with the geometry any response is likely to be speculation. You can upload a .3dm file by dragging it to where you type a post, or by using the vertical arrow icon above where you type a post.

1 Like

Hi Peter - I am not getting that here in my test - what is the scale? and the file tolerance?
Does SimplifyCrv before the Explode do anything good? And where it the curve seam on the circle relative to the rectangle?


File tolerance is 0.0003", the circle is 4"D. But here is a file.

bool.3dm (44.1 KB)

Hi Peter - thanks - so, here, CurveBoolean makes a curve that explodes into 4 segments, as expected - does that same exact file make something else for you?


Yes, that exact file. But, if you look at my 2nd video, you see that if I undo & then do it again, it explodes with 4, and if I undo & do it again, explodes with 6, etc. So try it, undo it, and try it again - a few times & see if you get 6.

Hi Peter - thanks - it looks so far like pick order matters - circle first = 6 segments, rectangle first = 4 segments. Thanks for pointing this out.

RH-58078 CurveBoolean pick order


Oh, interesting. I was using CTRL+A to select all. Thanks for identifying the problem!

Pascal, when I click on that YouTrack, page not found.

Also, is this the same bug that makes it so that when I trim a surface with a polyline, sometimes certain areas are merged into ‘one edge’ in the resulting trimmed surface and other times a different area is ‘one edge’? Super painful.