Thomas Bendall

Actually I'm not sure that there is necessarily a conflict between the uses for the intergrid kernels! The multi-grid solver in gungho currently only restricts and prolongs a W3 (fully-discontinuous)...

> 1. Looping limits change from `1, mesh%get_last_halo_cell(1)` to `1, mesh%get_last_edge_cell()` > 2. No halo exchanges are generated for readers in the kernel. > 3. No colouring is required in...

Ah sorry. I have some intergrid kernels that use stencils working on a branch and there wasn't any problem with the different halo depths. With respect to the original issue,...

Yes that's true, we have relaxed that constraint. I have to confess to not being able to remember where that assumption is made! It may be that we haven't noticed...

Yes, it does support varying nesting factors. I've reminded myself of what exactly it is that we're doing. With this kernel, we are writing to fine cells that are nested...

I thought I'd just add another comment here, as there is a whole category of kernels that currently have psykal lite implementations because they loop to some depth in the...

Thanks Andy, I agree with that summary and that it makes most sense to add this through the `operates_on` variable. I suppose that the invoke would also require a `depth`...

Thanks Andy, that sounds really good to me. Trying to think ahead, the depth to which one would loop into the haloes would probably be: - user defined to something...

Right now, I think there are only a handful of these kernels but they are pretty much evenly split between the above cases! Because there aren't many kernels, I'm not...

Hi Andy, yes that looks great, thanks. One final thought I have is about the description of `owned_cell_column`. If a kernel is performing an `INC` operation on a continuous field,...