[Core] Removing core warnings #12709
Conversation
| /// Components const Constructor | ||
| GeometryContainer( | ||
| GeometriesMapType const& NewGeometries) | ||
| : mGeometries(NewGeometries) | ||
| {} |
There was a problem hiding this comment.
Please someone else from @KratosMultiphysics/technical-committee confirm this is fine.
It is needed to call the const version of the WriteGeometries from the model_part_io.h
| rModelPart.Nodes().Sort(); // Avoid inadvertently reordering partway through | ||
|
|
||
| // Charlie: A write function should not modify the model part. This is a bug. If its broken needs to be fixed elsewhere. | ||
| // rModelPart.Nodes().Sort(); // Avoid inadvertently reordering partway through |
There was a problem hiding this comment.
@sunethwarna I think this fix is yours, we need to find another solution or call the sort outside this function.
There was a problem hiding this comment.
from my point of view this one is a blocker. This PR cannot go in until the PointerVectorSet is fully fixed
having said this i could accept a check here for the model_part.Nodes to be sorted. If it fails that is an error...
| // This needs to be somewhere else. | ||
| // r_nodes_array.Unique(); |
There was a problem hiding this comment.
Someone from the meshing can tall me why we need to reorder and remove duplicates from the modelpart here? A workaround would be to create a print modelpart but we cannot modify the original in a write function
There was a problem hiding this comment.
The problem is the operation or where the operation is placed?
There was a problem hiding this comment.
Where it is placed.
Right now is in a write method, which in the non deprecated IO needs a const & ModelPart, so we cannot do that. If the reorder is needed needs to be placed outside (which makes sense) but not sure where.
There was a problem hiding this comment.
The code you modified I don't think it will work. You cannot have const model part in the function that literally is retrieving the new mesh
There was a problem hiding this comment.
Yep that's my point, that function should not be in charge of doing that. Ideally you should have:
void foo() {
ModelPart my_model = GenerateModel()
Write(my_model)
}
but for some reason, in the meshing, the `GenerateModel()` part is done inside the write. I am trying to understand where this write code is being called to split it in two parts (the part that modifies the modelpart and the part that actually writes it)There was a problem hiding this comment.
but this one is needed for correctness, so i can go in only when all the ongoing changes of the PointerVectorSet (including the "authority" stuff are merged)
There was a problem hiding this comment.
including the "authority" stuff are merged
that's on me and will likely take ages
📝 Description
Gets rid of some annoying warnings in the core:
DivideInputToPartitionsandconstversions of functions inIO(more than 2 years being marked as deprecated)