I knew about these issues. Thank you for the animation.
For those who want to have endpoints side by side closer than SketchUp’s tolerance, 0,001", or avoid breaking a new edge as shown in the animation above, you can have them in different contexts. So in the aniation above, if the middle existing edge is in a different group, the new edge will not be split into two.
I’ll just throw this into the discussion.
These cubes are 1m, the pair on the right are correct, one raw, one group. The pair on the left have one vertex out by .005mm. You can see with the raw geometry you get an error that can increase with subsequent editing. And with the group it actually fails to see the face when creating the intersecting edges.
A couple of things that tend to limit the effect of this loose tolerance. Group/Component your geometry early and often, this can help to contain any errors in a repairable entity. Use the rectangle tool rather than the pencil tool wherever possible, a rectangle is less likely to get twisted than four individual edges. Keep an eye out for tiny edges at corners (a left to right selection of a corner should show nothing in entity info, if it shows an edge you know there is something in there)
Turn on hidden geometry regularly to spot any spidering and deal with it. And, as already mention, turn off length snapping. (I only repeat that here for those in the future that skim the thread)
Interesting demo.
That’s my go-to tool whenever possible. Can I trust that a face created this way is always 100% coplanar? Or can it still inference slightly off and generate a face?
I would frequently turn on hidden geometry when troubleshooting and never saw spiderwebbing with my slightly non-coplanar faces. Only by copying and pasting the face and intersecting faces was the fracturing visible. It would be helpful if Hidden Geometry had stricter tolerances than faces. I can’t think of any feedback SU gave me to alert me to my non-coplanar faces other than the ultimate face disappearance (probably due to repeated edits as per your demonstration). And I agree–disciplined construction techniques and the tips offered in the thread will drastically reduce such errors. Thanks for the input!
I don’t think I’ve been mentioned this much before!
To clarify one thing, what I was saying was that the bug where minute errors in edge locations can still leave you with apparently coplanar faces, is unlikely to get fixed. It’s down to us to notice those errors when they occur. They should not occur in any normal use, but certainly can come up more when importing CAD files.
So it is a “bug”? That’s actually something that wasn’t clear to me and is helpful to hear.
Up until now, the comments seemed to suggest (to me) that it is an unavoidable limitation of the programming and “computer arithmetic”. Or perhaps even a “feature” intended to make the program useable.
Regardless, I have a much increased understanding of how coplanarity works (or not) in SU, as well as how to find and avoid problems. The new tricks for rebuilding my model was a bonus (thanks for the slice-and-dice, Box!)
Thanks, also, for your input, Colin, and I’m sorry for my intensity–I meant no offense. For me, the better I understand how a program “thinks”, the better able I am to maximize its functionality and minimize my errors. It’s been a rough week (on multiple fronts) but I learned a lot. Now if I can just get through Tuesday–or technically December 14th–I might be able to taper off my meds!