[Core] Scalar/Vector variable assignment inconsistency

[matekelemen]

AssignVectorVariable*Process allows setting null to components of its value, but AssignScalarVariable*Process does not.

=> for example, if I want to constrain DoFs to whatever their state is at the time of constraining, I can't use AssignScalarVariable*Process.

Kratos/kratos/python_scripts/assign_vector_variable_process.py

Lines 73 to 82 in a4d6d95

	for indice,variable in enumerate(["_X", "_Y", "_Z"]):
	if not settings["value"][indice].IsNull():
	i_params = KratosMultiphysics.Parameters("{}")
	i_params.AddValue("model_part_name",settings["model_part_name"])
	i_params.AddEmptyValue("constrained").SetBool(settings["constrained"][indice].GetBool())
	i_params.AddValue("interval",settings["interval"])
	i_params.AddValue("value",settings["value"][indice])
	i_params.AddEmptyValue("variable_name").SetString(settings["variable_name"].GetString() + variable)
	i_params.AddValue("local_axes",settings["local_axes"])
	self.aux_processes.append( assign_scalar_variable_process.AssignScalarVariableProcess(Model, i_params) )

Kratos/kratos/python_scripts/assign_scalar_variable_process.py

Lines 79 to 90 in a4d6d95

	if settings["value"].IsNumber():
	self.value_is_numeric = True
	self.value = settings["value"].GetDouble()
	elif settings["value"].IsString():
	self.value_is_function = True
	self.function_string = settings["value"].GetString()
	self.aux_function = KratosMultiphysics.GenericFunctionUtility(self.function_string, settings["local_axes"])
	if self.aux_function.DependsOnSpace():
	self.cpp_apply_function_utility = KratosMultiphysics.ApplyFunctionToNodesUtility(self.model_part.Nodes, self.aux_function )
	else:
	self.table = ReadCsvTableUtility(settings["value"]).Read(self.model_part)

[loumalouomega]

I see, do you want to propose the change?

[sunethwarna]

What you say is true, in this case I use the fix_scalar_variable_process.py or fix_vector_variable_process.py
May be you can use them for the time being

[matekelemen]

I see, do you want to propose the change?

No, I'd rather spare myself the misery of babysitting a PR to core, if possible.

[RiccardoRossi]

I do not oppose to euxha change...

[rubenzorrilla]

I think that this was deliberate since the very beginning... I'm saying it by hearth, but as far as I remember, the vector process skips the component when setting it to null. Doing so in the scalar process would result in a do nothing process (this is why the null is not considered). I might be wrong, but this is why we created the "fix_wathever_variable_process.py", to always fix regardless the value.

[matekelemen]

If constrained has true components, it wouldn't skip it. In fact, it'd do exactly the same as "fix_wathever_variable_process.py".

Nope you're right, it doesn't do anything.

[rubenzorrilla]

Great. I think the issue is clear now and that we shall live with such inconsistency. @matekelemen should we close this or you prefer to keep it open?