BUG : Hidden corruption of surfaces

[HardRock]

Take a look at this sceenshots.
Part (shell) dispayed normal, but operation with this leads to incorrect result (for example, split apart with plane)

Выделение_133.png (20.38 KiB) Viewed 793 times

image_2020-10-17_221958.png (13.77 KiB) Viewed 793 times

If copy-pastle to another project, or export / import wiht STEP, part will shown broken

Выделение_134.png (25.92 KiB) Viewed 793 times

Originally, this part was imported from another project, then cloned and downscaled, after simple copy was created to lose history.
And it shows good, but internally it not good =)
Also if downgrade "broken" part to faces, the corruption will shown.

I have tried to reproduce all steps from original part, but it was not happens. So this is workaround


OS: Ubuntu 18.04.4 LTS (Unity:Unity7:ubuntu/unity)
Word size of OS: 64-bit
Word size of FreeCAD: 64-bit
Version: 0.19.
Build type: Release
Branch: unknown
Hash: df9282dc12d45996bb83cb42f6da10fd82519a47
Python version: 3.6.9
Qt version: 5.9.5
Coin version: 4.0.0a
OCC version: 7.3.0
Locale: Russian/Russia (ru_RU)

[HardRock]

If close and then open project file, faces shown corrupted. Seems like corruption was made "on the fly" in current session.
If this is interesting and you know how to make "snapshot" - i may try to do it.

After opening "broken file", repot has "The mesh data structure has some defects"

[openBrain]

Did the flaw on the surface also reveals when you force a full recompute?

[HardRock]

No, only after interraction, close / open / copy - paste or export import into another project.

Sorry i have accidentally closed buggy session
May be next time... Or I can attach good and broken parts if it may help to understand the reason.

[chrisb]

Moved from Open discussion.

What does Part->CheckGeometry say? You may attach a file for further inspection.

[HardRock]

Lot of errors.

In attachment "corrupted" and "good" (reassemled) parts

[HardRock]

Another corruption happens

After setting refine to true for cut operation, this face was corrupted irrevocably. It shown normally, but part check says that Edge1 is very small edge.
Using this face within shell will produce unexcpected results after operation on that shell (cut for example)

[OficineRobotica]

If i can add my testimony.... i'vd hit a very similar issue https://forum.freecadweb.org/viewtopic. ... 20#p423898
https://forum.freecadweb.org/viewtopic. ... 90#p432800
but this time in linkstage3 branch.
At first we thought it was a Boolean operation and then we tried to blame a fillet issue but now that you mention it I hit this kind of problem after refine on a feature.

This has happened on another model and always after a refine. After that the model is almost irecovreble if not by deleting allot of features prior to the refine seted one. It seems to happen with occ 7.4

Sorry for the confused report, I'm on mobile, but it seems a important issue. Unfortunatly it can cause allot of time loss when modelling

[HardRock]

Yes this is very bad bug that consume lot of time and happens often.

Mostly I have faced with this problem when shell vs solid operations. For example, cut solid from shell to make hole.

I think we should not use refine for that type of operations because this is not safe now and problem is not visible right after refine.
Also there is no way to rollback. Face corrupted without chance to get it back to "normal" state.

Current my workaround for this bug - remove corrupted face and make new one.

[Chris_G]

After setting refine to true for cut operation, this face was corrupted irrevocably. It shown normally, but part check says that Edge1 is very small edge.
Using this face within shell will produce unexcpected results after operation on that shell (cut for example)

Code: Select all

f1.getTolerance(1)
>>> 11.237659910743833

I don't know how it happened, but the Max Tolerance of this face is totally insane.