collection of Clip filter bugs

[wmayer]

This was my fix https://github.com/FreeCAD/FreeCAD/commit/ceb5a3da
but it does not work anymore.

It's not a good fix anyway because it's totally superfluous to regenerate the whole geometry after changing the transparency. Instead you only have to touch the face set node. This is because transparency is directly handled by SoIndexedFaceSet.

I bisected and this commit introduced the regression that the transparency for filters does no longer work for me:

The difficulty is to decide when to use the user requested transparency and when to use the transparency from the color bar settings. So, we could define that if transparency property is changed by the user (i.e a value > 0%) it will take precedence over the value coming from the color bar or alternatively the maximum value will be used.

But even when undoing the changes of git commit e437893e99 the transparency for the pipeline object is not fixed. I guess the problem is that the SoMaterial node isn't changed because when exporting the object in .iv format then the transparency value isn't modified, i.e. it's 0.

EDIT: The transparency is ignored if Field is set to None. This is true for pipeline and clipped objects.

[wmayer]

This was my fix https://github.com/FreeCAD/FreeCAD/commit/ceb5a3da
but it does not work anymore.

This caused the regression that if Field is set to None the transparency wasn't set. So, I reverted most of the changes and fixed it the right way.

I bisected and this commit introduced the regression that the transparency for filters does no longer work for me:

I have changed it to use the maximum transparency from user set transparency and from what the color bar delivers as this IMO is the most sensible way.

Then in ViewProviderFemPostObject::update3D() the range of the color bar was always overridden that made it impossible to set in the color bar and keep it like that. Now only changing the Field property will reset the color bar range again.

git commit 5822568566

What remains is the problem with the double-click on the one project file.

[wmayer]

What remains is the problem with the double-click on the one project file.

The odd thing with Electrostaticsplane-bug.FCStd is that the object is called SolverElmer but it uses the icon of the Calculix solver.

[wmayer]

For some reason the SolverElmer object lost its reference to the Python proxy object.

Code: Select all

App.ActiveDocument.SolverElmer.Proxy # is None

To fix it you have to:

• Open the file Mod/Fem/femsolver/elmer/solver.py
• Change the __init__ method of Proxy to

  Code: Select all

      def __init__(self, obj):
          super(Proxy, self).__init__(obj)
  

• Save the changes
• Start FreeCAD and load the project file
• Execute this Python code

  Code: Select all

  from femsolver.elmer.solver import Proxy
  App.ActiveDocument.SolverElmer.Proxy=Proxy(App.ActiveDocument.SolverElmer)
  from femsolver.elmer.solver import ViewProxy
  App.ActiveDocument.SolverElmer.ViewObject.Proxy=ViewProxy(App.ActiveDocument.SolverElmer.ViewObject)
  

• Undo the changes in solver.py

[uwestoehr]

This caused the regression that if Field is set to None the transparency wasn't set.

Hmm, it fixes that the transparency was at least set at all. The problem was that there was a loop between WriteTransparency() and update3D().

So, I reverted most of the changes and fixed it the right way.

Thanks! I can confirm it works now for filters and pipelines.

[uwestoehr]

To fix it you have to:...

The interesting point is why the issue happened. All my real-life simulations using Elmer are still OK. it only happened for the 2 exampe files I uploaded to GitHub.
Therefore I think, let's postpone this for now.

FC 0.20 is close to be ready, there is now only one known crash left in the surface WB and yesterday I triggered the LibPack development.

Many thanks for all your great work and patience with me

I am now away from my development PCs - holiday

[wmayer]

Hmm, it fixes that the transparency was at least set at all.

The way how it was before the transparency was set too but the rendering wasn't updated. Touching the SoIndexedFaceset node fixes it.

[wmayer]

The interesting point is why the issue happened.

In FEM there took place many restructures of the Python modules and you had a rather old project. When opening the Document.xml file then the Proxy property of the SolverElmer is saved like this:

Code: Select all

                <Property name="Proxy" type="App::PropertyPythonObject">
                    <Python value="bnVsbA==" encoded="yes" json="yes"/>
                </Property>

while when it's fixed it looks like this:

Code: Select all

                <Property name="Proxy" type="App::PropertyPythonObject">
                    <Python value="e30=" encoded="yes" module="femsolver.elmer.solver" class="Proxy"/>
                </Property>

It should be checked if with femtools/migrate_app.py old project files can be supported.

EDIT: It's not possible because the decoded values are None, "" or 1. So, at some point the information was lost.

[uwestoehr]

In FEM there took place many restructures of the Python modules and you had a rather old project.

Thanks. I see now that I opened my file with FC 0.19 and saved it. Then I reopened and saved with FC 0.20/master. This explains it.