ci: indentation failure - why!?

[consideRatio]

We error in the k8s 1.26 / main test, because the built pod can fail to startup because of an indentation error in singleuser.cmd.

But we don't have an indentation error in the provided config:

binderhub/helm-chart/binderhub/values.yaml

Lines 213 to 231 in 9cdae11

	singleuser:
	# start jupyterlab server *if available*
	# fallback on jupyter-notebook
	cmd:
	- python3
	- "-c"
	- |
	import os
	import sys
	try:
	import jupyterlab
	import jupyterlab.labapp
	major = int(jupyterlab.__version__.split(".", 1)[0])
	except Exception as e:
	print("Failed to import jupyterlab: {e}", file=sys.stderr)
	have_lab = False
	else:
	have_lab = major >= 3

Yet we have it in the pod based on k8s namespace report:

---

Is the singleuser.cmd processed in any way by us or helm making this happen? Is variable expansion involved adding a space or similar?

[manics]

It's fine on main
https://github.com/jupyterhub/binderhub/actions/runs/5045025696/jobs/9048789662
Does it happen everytime on #1697 ?

[consideRatio]

I think so, it mayyyyyy be related to rebuilding the image if its failing there, which makes chartpress do things it otherwise may not do.

[consideRatio]

Okay this is what goes wrong:

binderhub/testing/local-binder-k8s-hub/install-jupyterhub-chart

Lines 38 to 39 in 2b824a5

	with open(os.path.join(helm_chart, "binderhub", "values.yaml")) as values_in:
	yaml.safe_dump(yaml.safe_load(values_in)["jupyterhub"], tmp.file)

Apparently this file dump doesn't represent the original YAML content.

[consideRatio]

From the binderhub repo's folder, this returns correct indentation (with non-critical indentation mistakes in the original python code) - so reading data is fine.

    import os
    from ruamel import yaml

    with open(os.path.join("helm-chart", "binderhub", "values.yaml")) as values_in:
        read_values = yaml.safe_load(values_in)
    
    print(read_values["jupyterhub"]["singleuser"]["cmd"][2])

import os
import sys

try:
    import jupyterlab
    import jupyterlab.labapp
    major = int(jupyterlab.__version__.split(".", 1)[0])
except Exception as e:
    print("Failed to import jupyterlab: {e}", file=sys.stderr)
    have_lab = False
else:
    have_lab = major >= 3

if have_lab:
  # technically, we could accept another jupyter-server-based frontend
  print("Launching jupyter-lab", file=sys.stderr)
  exe = "jupyter-lab"
else:
  print("jupyter-lab not found, launching jupyter-notebook", file=sys.stderr)
  exe = "jupyter-notebook"

# launch the notebook server
os.execvp(exe, sys.argv)

[consideRatio]

Reading, writing, and then makes it reproduce the issue. Since helm is reading it later, I presume that the writing is to blame.

import os
from ruamel import yaml

with open(os.path.join("helm-chart", "binderhub", "values.yaml")) as values_in:
    read_values = yaml.safe_load(values_in)
print(read_values["jupyterhub"]["singleuser"]["cmd"][2])

with open("dump.yaml", "w") as values_out:
    yaml.safe_dump(read_values["jupyterhub"], values_out)

with open("dump.yaml") as values_in_again:
    read_values = yaml.safe_load(values_in_again)
print(read_values["singleuser"]["cmd"][2])