Nico Schlömer

Note also that gmsh calls "tag" what we'd rather refer to as a function, i.e., ```python mesh.cell_data["line"]["cell_tags"] = [-1, -2, -1, 0, -2] ``` You probably realize that already.

I wouldn't worry about MSH2.2 at _all_. So yeah, let's see if we can get representation 2 to work. Let's best start with those formats where it's native.

I like `cell_sets` and `point_sets` a lot since, in my view, these are the only actual _tags_: Each cell can have multiple, doesn't need to have any. This _should_ be...

> each element of a block always having the same tags Sorry to ask, you probably have explained that somewhere before: Are those gmsh 4.1 blocks always of the same...

Ah, Albany! I used to work with Andy Salinger when I was at Sandia. > Is there something similar for the parallel case? Not yet, I'm afraid. Your best bet...

I've never heard of either of the formats. Please post links to the specifications.

The mesh you attached is a super-simple triangle mesh and meshio has no problems reading it. You probably meant to upload something else.

Ah yes, okay. Well that still needs to be implemented. Feel free to PR. :smile_cat:

That fails, but it's a meshio issue. It should probably warn that it can't write cell_sets yet, and that one should convert it to int_data first.

Is `sets_to_int_data` not good enough then?