Jago Strong-Wright

Didn't mean to commit those Project.toml and Manifest changes, will revert now

I've spent some time trying to diagnose the GPU issue and it's quite strange, when I copy the test line by line into a REPL it doesn't cause the memory...

So I think that the tests generally are very close to the memory limit (which I think is around 1GB, not sure how that's being set). It looks like the...

> > So I think that the tests generally are very close to the memory limit (which I think is around 1GB, not sure how that's being set). It looks...

> I'm still wondering about this solution, which may both be _more general_ (since users can effect arbitrary changes to tracer tendencies, not just source/sink terms) and is less heavy...

I think we do need the change I made to the tracked fields so that tracers and auxiliary fields can be tracked properly although I don't like the `(field_name=nothing, )`setup...

Thanks! > It seems maybe that the function `get_mask` (or `nearest_cell_center`? If user facing we should come up with a good name) may nevertheless be a crucial part of any...

Sure, I'll sort that out later

Sorry this took so long, I've put the example together and moved some functions to a different file in the source code. Do you think these are the appropriate functions...

Apologies, currently using `v0.76.1`. I can try to update if you think that could solve it?