Fusion of two adjacent cones fails

[plgarcia]

HEllo,
I found a bug, I reported in the bugtracker. I did not find this problem reported or dicussed.
Here is the link : https://www.freecadweb.org/tracker/view.php?id=3133


In short two cones parralel and identical can not be merged in one object.

I provide a minimal example: 2 identical cones and a fusion of these 2 cones that is not working. If you change the Z position of one of the cone almost whatever value (add a micron to it and make it -21.001), and then the fusion works. The fusion works as long as both of the ends of one cone are not in the same plan as one of the end of the second cone.
Some values to make more tests:
HC1 = Height cone 1
HC2 = Height cone 2
ZP1 = Z pos cone 1
ZP2 = Z pos cone 2

HC1 ZP1 HC2 ZP2
20 -21 20 -21 Failed end 1 cone 1 in the same plane as end 1 of cone 2, the same for ends 2
20 -21 19 -21 Failed end 1 cone 1 in the same plane as end 1 of cone 2
20 -21 19 -20 Failed end 2 cone 1 in the same plane as end 2 of cone 2
20 -21 20 -20 Works

I think this information suffices to investigate and correct the problem encountered.
You can also note that the icon link to an object with errors is not restored when opening the saved file.

Regards

[Kunda1]

I found a bug, I reported in the bugtracker. I did not find this problem reported or dicussed.

That bug has been closed per @normandc's response

In bugCone1.FCStd, the document needs to be recomputed for the fusion to fail.

The reason why it fails is simple: the two cone bases are tangent to each other and only meet on one infinitesimal point. This is an invalid condition because it would create a non-manifold solid.

Therefore this bug is invalid and should be closed as such.

[DeepSOIC]

btw, no failure here:

OS: Windows 8
Word size of OS: 64-bit
Word size of FreeCAD: 64-bit
Version: 0.17.11018 (Git)
Build type: Release
Branch: master
Hash: 0ad2e47f4d611a22f679e5b3ef5ed6851f8a02be
Python version: 2.7.8
Qt version: 4.8.7
Coin version: 4.0.0a
OCC version: 7.1.0

The fusion result is a compound of two solids, which share a vertex where these cones touch. The result passes geometry check.

[plgarcia]

Thanks,
@NormandC : there is no functional reason the fusion of 2 objects whatever they are could not be fused, except cases not forseen by the programmer.

Il is than solved in solved in the 0.17 version.

I will be waiting.

Thank you very much
Pascal Garcia

[chrisb]

It works here as well:
OS: Mac OS X
Word size of OS: 64-bit
Word size of FreeCAD: 64-bit
Version: 0.17.11614 (Git)
Build type: Release
Branch: (HEAD detached at a8faef0)
Hash: a8faef009b2e146767da8310ccc596c7ee63bb83
Python version: 2.7.13
Qt version: 5.6.2
Coin version: 4.0.0a
OCC version: 7.1.0
Locale: German/Germany (de_DE)

while it fails in the 0.16 stable:

OS: Mac OS X
Word size of OS: 64-bit
Word size of FreeCAD: 64-bit
Version: 0.16.6706 (Git)
Build type: Release
Branch: tag: 0.16.6706
Hash: f86a4e411ff7848dea98d7242f43b7774bee8fa0
Python version: 2.7.11
Qt version: 4.8.7
Coin version: 3.1.3
OCC version: 6.8.0.oce-0.17

[NormandC]

@NormandC : there is no functional reason the fusion of 2 objects whatever they are could not be fused

Yes there is: common sense.

FreeCAD is not about creating theoretical mathematics models. CAD is by definition about creating models intended to be made in real life. Two cones only joined by a vertex are impossible to make or the joint would break as soon as you took them into your hands. This is why valid models need to be manifold ("watertight").

If you want to group them to move as one, then it's a Part Compound you should use. A Boolean operation is about parts that intersect each other.

The proof is that if you check the succeeding Fusion in 0.17, it is actually a compound of two solids.

FC017_fused_cones_but_not_really_fused_01.png (13.71 KiB) Viewed 2475 times

btw, no failure here:

I consider this a fluke. I would bet money that in the long run, on some future OCC version, it won't succeed anymore. As tanderson69 once said ironically (from memory, I don't have the exact words): for a B-rep manifold kernel, OCC has a great capacity to generate non-manifold models (when it shouldn't).

[plgarcia]

I understand these arguments of object to be made in real life. I my case this object is an intermediate object that will be subtracted to another object and make holes. The final result is the real object and can be made.

This problem of object that can be mede or not is to be addressed later on CAM phase.

I think that all operation should be accepted when they make sense, at least between two solids. Solid could always be added, intersect, subtracted, and produce a solid and eventually remove all part of the result that is not solid such as isolated curves, lines, dots if it can cause problem for further operations.

Furthermore the user should be given more help when a problem occurs.
Regards
Pascal Garcia

[chrisb]

urthermore the user should be given more help when a problem occurs.

Yes!

Currently high quality additional help is found here in the forum.

[plgarcia]

Sorry, I was not saying the help the forum provide is not good, and I thanks everybody having taking time to help.
A flying Windows appears saying “Fusion failed”. It would have been nice to be given a bit more explanation on the reason the errors occurs, eventually an error code that could be documented somewhere else.
Regards

[plgarcia]

The issue has been closed.
Nevetheless, the problem has dissapear in the future 0.17, so I beleive a modification has been introduced at a certain time that solve this problem and may be some others.
According one comment in the issue it was due to the fact that the two cones were tangent and intersection was only a point, that obvously was not properly handled by the software.
Regards
Pascal Garcia