================================================
This should fix this (I cannot reproduce it):
https://forum.freecad.org/viewtopic.php?p=667579#p667579
Lately I added the geometry pointer to the item. This has indeed
the potential for an already deleted pointer being accessed.
This PR removes the geometry pointer from the item and relies on the
ViewProvider to indirectly access an updated pointer.
====================================================================================
As reported:
https://forum.freecad.org/viewtopic.php?p=667426#p667426
Support for moving external geometry to another layer will come in the future.
This commit adds the ability to shift the involute profile inside or
outside. Profile shift is implemented as coefficient, i.e. normalized by
the module, so that it the whole profile scales with the module without
chaning shape.
To verify the profile, the tests implement an "over pins measurement"
using formulae found in literature.
Backward compatibility with FreeCAD-v0.20 is garanteed by already
existing tests, not touched by this commit.
This addresses issue #5618.
- the heat capacity of water was wrong
- also specify just thermal expansion, this make that water is recognized as material and then all water parameters will be available
(that the analysis fail despite CalculiX reports everything went well s is another issue, I will investigate later)
- the density must be output using the correct units. Due to this bug also the viscosity was written with wrong units-
- if there is a dynamic viscosity given for a material, use it directly
* in QtConcurrent::run the order of 1st and 2nd argument are swapped
* QtConcurrent::run asserts that the argument of the passed function pointer is not non-const
* Used methods of QFontDatabase are static in Qt5 and Qt6
* QTextStream::setCodec() has been removed in Qt6
* Argument of enterEvent() has changed from QEvent to QEnterEvent
* QTextCharFormat::setFamily() is deprecated
- it is a long-standing bug that the name of the result mesh object is not unique. For example for a frequency analysis you get for every result the same object "ResultMesh". When now a "ResultMesh" object is deleted because one result mode object should be deleted, all result modes loose their mesh.
To fix this, use a unique name for the mesh object.
- make the connection also working for frequency and buckling analysis by directly creating/updating the pipeline where the CalculiX results are loaded
