fix: Get CI running again

[matthewfeickert]

I saw that jupytext's CI was broken so thought I'd try to help fix it. There's lots of cleaning that could be done on the CI, but this just gets it running again.

This updates all the GitHub Actions to their latest releases and updates and applies the pre-commit hooks so those pass as well. It additionally fixes the dependency conflicts happening by trying to install the linting dependencies for older Pythons by just having the lint job be separate as it only needs to run once, not for every Python being tested.

I also threw in the use of python -m pip pattern over just pip even though it isn't needed to try to spread this best practice to others who might be looking at jupytext's CI. Feel free to ask for that to be removed if you aren't a fan.

[matthewfeickert]

@mwouts this is ready for review.

[mwouts]

Wow thank you so much @matthewfeickert , this is very helpful! I will have a look at this asap.

[mwouts]

This looks really great! Thanks again.

I see in your comment above that more cleaning could be done, that could be a great idea, feel free to tell us more about that!

[matthewfeickert]

I see in your comment above that more cleaning could be done, that could be a great idea, feel free to tell us more about that!

Sure. Some of this is just matters of preference, and obviously whatever works best for you is what you should do, but I've found it easier to maintain GitHub Actions workflows by separating them out into multiple workflows rather than having them all exist in one, like you normally would do for Travis CI, GitLab pipelines, etc. So for example, the

jupytext/.github/workflows/continuous-integration.yml

Lines 58 to 85 in e097422

	analyse:
	name: Analyse
	needs: skip_duplicate
	if: ${{ needs.skip_duplicate.outputs.should_skip == 'false' }}
	runs-on: ubuntu-latest
	steps:
	- name: Checkout repository
	uses: actions/checkout@v3
	with:
	# We must fetch at least the immediate parents so that if this is
	# a pull request then we can check out the head.
	fetch-depth: 2
	# If this run was triggered by a pull request event, then checkout
	# the head of the pull request instead of the merge commit.
	- run: git checkout HEAD^2
	if: ${{ github.event_name == 'pull_request' }}
	# Initializes the CodeQL tools for scanning.
	- name: Initialize CodeQL
	uses: github/codeql-action/init@v1
	# Override language selection
	with:
	languages: python, javascript
	- name: Perform CodeQL Analysis
	uses: github/codeql-action/analyze@v1

job doesn't need to live in your CI workflow, but could exist as a separate codeql-analysis.yml workflow.

Also, unless I'm mistaken about its purpose, the

jupytext/.github/workflows/continuous-integration.yml

Lines 11 to 19 in e097422

	skip_duplicate:
	runs-on: ubuntu-latest
	outputs:
	should_skip: ${{ steps.skip_check.outputs.should_skip }}
	steps:
	- id: skip_check
	uses: fkirc/skip-duplicate-actions@master
	with:
	github_token: ${{ github.token }}

can be replaced with GitHub Actions concurrency control (c.f. https://docs.github.com/en/actions/using-jobs/using-concurrency). i.e. something like

concurrency:
  group: ${{ github.workflow }}-${{ github.ref }}
  cancel-in-progress: true

These are just some thoughts, but more discussion would probably be better done in an Issue. I'll make one from this comment.