mesh_from_numbers needed these additional lines in cell 6 to run for the UM Tutorial in Melbourne

[gryffydd]

mesh_from_numbers needed these additional lines in cell 6 to run for the UM Tutorial in Melbourne

import sys
sys.path.append('/g/data/access/ngm/data/umtutorial2023/ngux/iris-mesh-tutorial/notebooks/')

import pyvista
pyvista.start_xvfb()

[pp-mo]

Thanks !

I'm not sure of the facts here, and if this is the best way. Can you comment @davelee2804 ?
I imagine we should fix this by creating a PR against the 'NCI_ARE' branch.

[pp-mo]

if this is the best way

In particular, the pyvista.start_xvfb() .

From a quick look at the differences between the 'NCI_ARE' and 'main' branches, I see that the core change is to add this into all notebooks.
I suspect that we should maybe change how this works by fixing the default backend.
I also think that xvfb doesn't exist in the env spec, or the locked env, so this could/should be fixed in the 'NCI_ARE' branch, at least if that is expected to have a more than a transient existence ?

For future, if NCI_ARE is to persist, we may need to think about how it relates to 'main' + how we would stop them drifting apart. I'm just thinking perhaps there is a better way of making the differences "switchable" within a common codebase.

[gryffydd]

This was related to the NCI_ARE branch and I apologise for not being clearer.

I was not sure if this was the right way to report the issue on that branch

[davelee2804]

Hi all, we really just put the NCI_ARE branch together quickly so that we could have something running for the workshop last Friday. The pyvista.start_xvbf() call was necessary to fix a DISPLAY issue that we were experiencing. From memory I think I got this from a previous error message in some GeoVista notebooks I had experimented with a few weeks earlier.