Open domfournier opened 6 months ago
Interesting, there must be some missing logic wrapping around negative levels when not all cell width arrays have the same length.
Also… why are you using negative cell widths?
We should probably make the meshes error on this case due to undefined behavior.
Came about while testing the conversion between the octrees stored in geoh5 from and to discretize. GA can deal with either positive or negative cell sizes , so it made sense to check.
I can just block the conversion if it happens for now - no big deal. Just thought I would flat it.
This minimal example demonstrates the issue. The cell state appears not preserved if the cell size is negative.