So... is @ppemawm using different OCC in his Assembly4 model?
or what is making his files different/smaller?
So... is @ppemawm using different OCC in his Assembly4 model?
Thank you for these explanations.realthunder wrote: ↑Sun Jun 05, 2022 8:18 am It seems to be an OCC problem. I got the file you privately sent to me, and extract it (because FCStd is a zip file in case you don't know). Just sort the extracted files by size, you'll see the largest file is Fillet.Shape.brp (i.e. the shape of the fillet you created) being at 147MiB, and the second largest is Body003 at 139MiB, which is basically just a copy of Fillet. Don't know why it is a bit smaller, but not that much of difference anyway. The third largest is Boolean037 at only 34.9MiB. Note that those are uncompressed size, while FCStd is a compressed zip archive.
So I dig a bit deeper, and found out that indeed those new faces generated by the fillet are responsible for the bulk of the file. The following code selects the faces of the fillet that are bigger than 10MiB.
Sorry I didn't understand your "pb", sorry problem!
PartDesign in my branch supports multiple solids, distinguished by different colors in the tag icon attached to each feature. The problem you have is due to the handling of the Boolean feature, which usually produce a solid feature, and the body will treat it as a continuation of the current solid group. However, in your case, it produces only edges. The chain of solid features is broken. The current logic of the body is to assume that any new feature based on this non solid Boolean starts a new solid group, which is why it is not merged with the previous solid feature. I'll see how to handle these type cases better.jpg87 wrote: ↑Sun Jun 05, 2022 9:57 am Thank you for these explanations.
Another detail, I thought it was impossible to make several stages of the construction tree visible simultaneously!
Finally another question, did I make a mistake when I said that the steps that follow unSweep in my construction are not united with the rest?
When I open it either in the latest Linkstage3 or Master snap I get the following message:realthunder wrote: ↑Tue Jun 07, 2022 12:21 pm I have simplify the project to show OCC geometry size problem.
Code: Select all
15:20:39 AttachEngine3D::calculateAttachedPlacement: path curve second derivative is below 1e-14, can't align x axis.
Thank you, it's always a pleasure to have your explanations.realthunder wrote: ↑Tue Jun 07, 2022 12:21 pm I have simplify the project to show OCC geometry size problem. ...
Code: Select all
OS: Windows 10 Version 2009
Word size of FreeCAD: 64-bit
Version: 2022.430.26244 +4758 (Git)
Build type: Release
Branch: LinkDaily
Hash: b024b876480a533018fdc702145e9e38f8c54ba6
Python version: 3.8.6+
Qt version: 5.15.2
Coin version: 4.0.1
OCC version: 7.6.0
Locale: French/France (fr_FR)
Code: Select all
OS: Windows 10 (10.0)
Word size of FreeCAD: 64-bit
Version: 0.20.29045 (Git)
Build type: Release
Branch: master
Hash: af811b34b3
Python 3.8.13, Qt 5.12.9, Coin 4.0.0, OCC 7.5.3
Locale: French/France (fr_FR)
Installed mods:
* Assembly3 0.11.3
* BOLTSFC
* Curves 0.4.3
* DynamicData 2.46.0
* fasteners 0.3.41
* fcgear 1.0.0
* PieMenu
* symbols_library
Do you know if there has been a bug filed in occt for this?realthunder wrote: ↑Tue Jun 07, 2022 12:21 pm I have simplify the project to show OCC geometry size problem. The project can be opened in upstream master. This tiny file will explode when you make a fillet for the edge shown in the picture below. The three modified faces are all around 45MiB each. We can export that fillet shape as STEP file and open it to find more clues. Each of those faces are bounded by BSpline curves with about 65K control points! This surely indicates some problem in OCC.
Not that I know of. Maybe you can look into it? Because of the discovery here, I begin to notice that occt auto generated BSpline Curves/Surfaces are generally big, commonly exceeding thousands of control points, and sometimes with degree of one (i.e. all line segments). Not sure if it's always like that or just in recent versions.tanderson69 wrote: ↑Sat Jul 16, 2022 5:07 pmDo you know if there has been a bug filed in occt for this?realthunder wrote: ↑Tue Jun 07, 2022 12:21 pm I have simplify the project to show OCC geometry size problem. The project can be opened in upstream master. This tiny file will explode when you make a fillet for the edge shown in the picture below. The three modified faces are all around 45MiB each. We can export that fillet shape as STEP file and open it to find more clues. Each of those faces are bounded by BSpline curves with about 65K control points! This surely indicates some problem in OCC.