Install Tox at system level

[johnsca]

Not all runners will have ~/.local/bin on the PATH by default, so Tox should be installed at the system level.

[Cynerva]

Any idea what's up with the failing microk8s and microstack tests? Are they expected to fail right now? It looks like they're unrelated to the changes in this PR but I have no idea what's going on.

[johnsca]

Even though the microstack case is still failing, since it's used less than LXD or microk8s, I'd like to get this merged and those fixed and merged and we can work on the microstack issue going forward.

[DomFleischmann]

Should we maybe add a basic pytest-operator test to the testing actions which builds and deploys some charm with charmcraft? This might help with some of these issues in the future.

[DomFleischmann]

When trying to run an action with this branch I get the following error:

  /usr/bin/bash -c sudo usermod -a -G microk8s $USER
  /usr/bin/sg microk8s -c scripts/microk8s-init.sh
  sh: 1: scripts/microk8s-init.sh: not found
  Error: The process '/usr/bin/sg' failed with exit code 127

[vicamo]

Besides, b12e61c added new conditions for branch main, and yet main is not even exist in this repository. As result pull requests to master will never trigger CI.

[johnsca]

@vicamo Thanks, I meant to rename the primary branch and missed that. Fixed now.

Edit: Actually, e673f54 was also included as a stop-gap, but not reason to not go ahead and do the rename.

[vicamo]

@vicamo Thanks, I meant to rename the primary branch and missed that. Fixed now.

Edit: Actually, e673f54 was also included as a stop-gap, but not reason to not go ahead and do the rename.

Hi, thank you for picking this up. But actually deleting master branch fails all existing downstream projects, if any. I think maybe we'd prefer a more considerate way, taking advantages of revisions, to prevent breaking people's projects unsolicitedly. After all, this project is meant for wide use, isn't it?