Hi Michael,

Can you please explain in the .3dm file attached when I extrude the shape in green picking both profiles why it doesn't behave like the square profiles I want a narrow wall.

-

The Boolean difference step seems to be not working.

The inner solid is there, and can be Boolean differenced away from the outer solid, to create the shell.

- Brian

Hi Danny, it seems they are at some teeny tiny little difference in elevation from one another. Maybe just in one little segment or something, I haven't been able to visually see where the deviation is but if you go to the Front view and grab the corner of the edit frame and bring it down you will get a "flat" snap which won't happen unless there is something to flatten...

Anyway after doing the flattening from the elevation side view, the extrude will then behave how you were expecting, I've attached an updated file here.

- Michael

Hi Danny, after a bit more investigation your curves are only different in elevation by a very tiny amount, like 1e-12 , they are so close enough to the same that it should be ok to use them as-is, but the curve sorter mechanism was using a too tight tolerance and deciding that they should be extruded separately from one another.

I've tuned this up for the next v3 beta so that the curve sorter will use a more reasonable tolerance for coplanarity detection and that will make your original file work in the next v3 beta. Thanks for posting the example file!

- Michael

Thanks Michael.

-

It would be nice to have some type of option though, to know this when it is the case. I use MoI a lot to "fix" bad geometry. I would like to have a way to turn on "I want to see when they are out by even a bunnies hair" mode.... I don't always want, "Loosen it up so that it looks ok".......

Is that possible? = MoI Modes = "Free and easy" or "so tight the butter doesn't leave the biscuit" mode....

Hi Burr, well the problem is that in this case the amount that things are off was by 0.0000000000001 units, it's so small that it's quite hard to see it. Just seeing that extrude behaves differently isn't really a great diagnostic tool itself, it would be better to later on have more specific diagnostic tools that were focused on doing actual analysis.... Rather than just a mode for making operations fail I mean like I guess is basically what you're asking for?

- Michael

Yes. The fact that the extrude fails (Or is open) would have me immediately go for flat snap. If it completes, I'd pass on the bad geometry. But for the "not a high end modeler, but easy and not cluttered and fun", I could see all these requests I have fall into the "no need" Category.

If you have some idea of your future analysis "extended properties" ui that may handle this, then I can look forward to that.

But I hope you see my point about the fact that it failed being an "information point" for me. It didn't tell me details about "what", but I can make educated decisions from that point.

I guess it also depends on the output from the new change. Does the new extrude have a "slightly out" area? Or does it now think it's planar? That's where the cluttered other UI's would pop something out into your face and ask you which of 4 ways you would like to handle the extrude, I guess.

Hi Burr,

> If you have some idea of your future analysis "extended properties" ui that may handle this,
> then I can look forward to that.

Yup, I do have some ideas along those lines for the future.


> But I hope you see my point about the fact that it failed being an "information point" for
> me. It didn't tell me details about "what", but I can make educated decisions from that point.

Sure, I do understand that. But it's just not really practical for me to focus on keeping modeling operation failures around for a kind of "side effect" like that...



> Does the new extrude have a "slightly out" area?

The new extrude does have a "slightly out" area but the amount of it is so extremely minimal (around 0.0000000000001 units...) that I don't think it can really make any difference in any practical way... There are plenty of other things like general surface/surface intersection that are going to end up looser than that anyway. If an error of that tiny amount is not acceptable you would probably have to build things only out of planes and cylinders and nothing else.


> Or does it now think it's planar?

It will now consider these 2 to be coplanar. It did think that they were each planar before, but not that they were on the same plane.

- Michael

"""""""""""But it's just not really practical for me to focus on keeping modeling operation failures around for a kind of "side effect" like that..."""""""""""""""

But it wasn't really a failure. I mean, if the curves are not planar, then the extrude is an open surface. That's just the way I understand MoI.

The fact that the original 2d geometry is out by only .000000000000000001 is more incidental to the fact that it's "out"... For me.

Small little "rotations" in geometry will produce errors in operations further down the road in my model, that will make it much harder to get back and fix. Things like "How the heck did those 2 surface seams get just far enough off to produce a teeny tiny little fragmented sliver that causes all fillets and further Booleans to fail in that area, 3 days after I originally made that mistake... Kind of thing.

I understand what your saying, I just wanted to convey the importance to me, to you, about keeping an ability to catch those..... :)

Hi Burr,

> Small little "rotations" in geometry will produce errors in operations further down the
> road in my model, that will make it much harder to get back and fix.

Yeah that's certainly true that misalignments can cause problems later on.

But at a certain point when they get small enough it's a different kind of thing - if MoI only considered things that were "100% exactly" on the same plane as being planar, a whole lot of things would stop working at all since it's just fundamental to the way floating point mathematics on the computer work that there is a teeny tiny amount of fuzzy-ness to coordinates since the computer has to represent numbers in a limited amount of memory.

Don't get me wrong - it's not bad in general to want to have accuracy but it's just part of the whole way CAD on the computer works that there is a tiny amount of fuzz involved, and when the fuzz is teeny tiny small enough (like we're talking about in this specific case) it's a normal and expected thing.

Misalignments have to be a fair amount larger than 0.0000000000001 units for them to cause problems, generally speaking.

- Michael

Hi Michael,

> It will now consider these 2 to be coplanar. It did think that they were each planar before, but not that they were on the same plane.

That's interesting, so the 2 profiles were planar, I made the inner profile from circles then Boolean Union them together applied fillets then I took the whole profile and offset it outward, the only thing I can think of to check is the centres that I used for reference.

0.0000000000001 would be quite acceptable at work for precise Engineering seeing that our CAD analysis tools are set to show 9 decimal places anything after that wouldn't be seen and never had a problem even if we had a result of 1e-9 we wouldn't worry about it.

-

"""" it's just fundamental to the way floating point mathematics on the computer work that there is a teeny tiny amount of fuzzy-ness to coordinates """"""""

Ok, understand.....