0.20 bug/regression Array Pattern issue #6641

About the development of the Part Design module/workbench. PLEASE DO NOT POST HELP REQUESTS HERE!
jnxd
Posts: 576
Joined: Mon Mar 30, 2015 2:30 pm

Re: 0.20 bug/regression Array Pattern issue #6641

Post by jnxd »

wmayer wrote: Tue Apr 12, 2022 2:34 pm git commit 5ab7d17f67
Now it's your turn.
https://github.com/FreeCAD/FreeCAD/pull/6738
user1234
Veteran
Posts: 1675
Joined: Mon Jul 11, 2016 5:08 pm

Re: 0.20 bug/regression Array Pattern issue #6641

Post by user1234 »

Sorry for being annoying, but i have tested the PR and other stuff and i have remarks and many many question.

Code: Select all

OS: Debian GNU/Linux 11 (bullseye) (X-Cinnamon/lightdm-xsession)
Word size of FreeCAD: 64-bit
Version: 0.20.28691 (Git)
Build type: Debug
Branch: wmayer_issue_6641
Hash: 2871aa3823f1b49039f83b92a2102a030f99c317
Python 3.9.2, Qt 5.15.2, Coin 4.0.0, OCC 7.4.0
Locale: English/United States (en_US)
(except, when i write something different for testing)



First off all, for the most normal operations (rectangle, round, plane) the transform tool works, but sometimes with more complex shapes i found some funny behaviors.

I managed in an old test build before, that all fillets and transform work in the file in the linked post, filename: Kappe50-jeno_edit.FCStd.

https://forum.freecad.org/viewtopic.php ... 40#p582840

This was OCCT7.4.0 with git commit 89b8e81d5c3b11572a0288720d0b4145d2ec2a7e


Now with retesting it with the same OCCT and commit, it do not work any more and i do not know why. Unfortunately while cleaning yesterday, i deleted the original build.



When i tested the PR, originally this also do not work. Then while clicking around the properties, i found out, there is property called Support Transform for features like fillets, chamfers, draft and thickness.

When i set it to true on Fillet001, then it works, but i did never changed it before, also on the test file (when you open it, you will see it).

PR1.png
PR1.png (230.51 KiB) Viewed 947 times
PR2.png
PR2.png (114.56 KiB) Viewed 947 times


When i also set it true on Fillet, then the transformation succeed, but the result is unexpected, because the transformation of the Fillet is not performed, which his also for the name of the property misleading.

PR3.png
PR3.png (217.92 KiB) Viewed 947 times
PR4.png
PR4.png (225.01 KiB) Viewed 947 times


So my question: what does the property? That is the purpose of it? I found nothing about that. Is the property incorrect implemented or normally deprecated all all?


Also in the PR brach version, the bace face do not refine, even when Refine = true. In this version, OCCT7.4.0 with git commit 89b8e81d5c3b11572a0288720d0b4145d2ec2a7e, it does.

Just also a remark, it no regression against 0.19, there this do not work at all. It is a regression about a dev-build. I do not know (and i do not rate) if it has an impact about this evaluation.


Fun fact: with OCCT7.6.1 the transformation do not work with the testfile.

Greetings
user1234
wmayer
Founder
Posts: 18776
Joined: Thu Feb 19, 2009 10:32 am

Re: 0.20 bug/regression Array Pattern issue #6641

Post by wmayer »

user1234 wrote: Tue Apr 12, 2022 10:11 pm Now with retesting it with the same OCCT and commit, it do not work any more and i do not know why. Unfortunately while cleaning yesterday, i deleted the original build.
I have tested Kappe50-jeno_edit.FCStd with current master build that includes my changes from yesterday. After opening the file I made a simple copy of PolarPattern and then recomputed it.

The only thing I realized is that the recomputed feature looks a bit smoother because of more triangles which you can see visually when activating the render mode "Hidden Lines". The reason for different number of triangles is I guess the different OCC version -- mine is 7.3.0.

However, the actual shape geometry is still the same. In order to find out any differences I switched to the Inspection workbench and made a visual inspection between the copied and recomputed polar pattern. In order to see any tiny differences in the color plot I had to change the range to [-5e-6, +5e-6], i.e. 5 microns.
So my question: what does the property? That is the purpose of it? I found nothing about that. Is the property incorrect implemented or normally deprecated all all?
Tooltip says:
Include the base additive/subtractive shape when used in pattern features.
If disabled, only the dressed part of the shape is used for patterning.
Inside the code it's used here https://github.com/FreeCAD/FreeCAD/blob ... p.cpp#L227 but its purpose is not totally clear to me.
user1234
Veteran
Posts: 1675
Joined: Mon Jul 11, 2016 5:08 pm

Re: 0.20 bug/regression Array Pattern issue #6641

Post by user1234 »

wmayer wrote: Wed Apr 13, 2022 3:07 pm I have tested Kappe50-jeno_edit.FCStd with current master build that includes my changes from yesterday. After opening the file I made a simple copy of PolarPattern and then recomputed it.

Does not work here:

Code: Select all

OS: Debian GNU/Linux 11 (bullseye) (X-Cinnamon/lightdm-xsession)
Word size of FreeCAD: 64-bit
Version: 0.20.28717 (Git)
Build type: Release
Branch: master
Hash: cfe94b00d3434f3e2c8b5114d2909f628c330fa8
Python 3.9.2, Qt 5.15.2, Coin 4.0.0, OCC 7.6.1
Locale: English/United States (en_US)
and OCCT7.5.3 and 7.4.0.


The result is only a small fillet surface, the rest is vanished.

master_with_OCCT7.6.1.png
master_with_OCCT7.6.1.png (268.92 KiB) Viewed 727 times

master_with_OCCT7.5.3.png
master_with_OCCT7.5.3.png (274.45 KiB) Viewed 727 times

master_with_OCCT7.4.0.png
master_with_OCCT7.4.0.png (316.89 KiB) Viewed 727 times


wmayer wrote: Wed Apr 13, 2022 3:07 pm Tooltip says:

Include the base additive/subtractive shape when used in pattern features.
If disabled, only the dressed part of the shape is used for patterning.

Inside the code it's used here https://github.com/FreeCAD/FreeCAD/blob ... p.cpp#L227 but its purpose is not totally clear to me.
This is why i ask, because the beavior is not like it should be. It seems like a depricaded property.



Greetings
user1234
wmayer
Founder
Posts: 18776
Joined: Thu Feb 19, 2009 10:32 am

Re: 0.20 bug/regression Array Pattern issue #6641

Post by wmayer »

The result is only a small fillet surface, the rest is vanished.
With a OCC 7.6.0 build it fails to recompute the polar pattern. The error message is: Boolean operation failed
jnxd
Posts: 576
Joined: Mon Mar 30, 2015 2:30 pm

Re: 0.20 bug/regression Array Pattern issue #6641

Post by jnxd »

user1234 wrote: Sat Apr 16, 2022 3:15 pm
This is why i ask, because the beavior is not like it should be. It seems like a depricaded property.
If the result is different based on the property, it can't be deprecated.
user1234
Veteran
Posts: 1675
Joined: Mon Jul 11, 2016 5:08 pm

Re: 0.20 bug/regression Array Pattern issue #6641

Post by user1234 »

jnxd wrote: Sun Apr 17, 2022 5:48 am If the result is different based on the property, it can't be deprecated.
But since most of the time it works anyway and the result is sometimes inverted, is has no usecase more. Maybe stall is a better term.

Greetings
user1234
jnxd
Posts: 576
Joined: Mon Mar 30, 2015 2:30 pm

Re: 0.20 bug/regression Array Pattern issue #6641

Post by jnxd »

user1234 wrote: Sat Apr 23, 2022 4:55 pm Maybe stall is a better term.
Not sure of the term but yes if there's an adverse effect by the property it can go.
Post Reply