Cancel prev runs for bench and nix

[jneira]

• It is working really well to the test workflow

[pepeiborra]

If I understand correctly this is a new feature of Github CI providing built-in support to deal with duplicate actions.
So far we have been using a third-party action to do this (fkirc/skip-duplicate-actions).

It is working really well to the test workflow

It looks like the test workflow now has both mechanisms enabled, is that what we want?

haskell-language-server/.github/workflows/test.yml

Lines 24 to 28 in 9fa6abe

	- id: skip_check
	uses: fkirc/skip-duplicate-actions@v3.4.0
	with:
	cancel_others: false
	paths_ignore: '["**/docs/**", "**.md", "**/LICENSE", "install/**", "**.nix", "flake.lock", "**/README.md", "FUNDING.yml", ".circleci/**"]'

haskell-language-server/.github/workflows/test.yml

Lines 8 to 10 in 9fa6abe

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

[jneira]

It looks like the test workflow now has both mechanisms enabled, is that what we want?

Oh i forgot to set it to false as i did for the test workflow, thanks. However we are using other features of skip-duplicate-actions so i would keep it around (like in test workflow)

[pepeiborra]

It looks like the test workflow now has both mechanisms enabled, is that what we want?

Oh i forgot to set it to false as i did for the test workflow, thanks. However we are using other features of skip-duplicate-actions so i would keep it around (like in test workflow)

Can you elaborate on why we need both?

[jneira]

Can you elaborate on why we need both?

Cause we are using skip-duplicate-actions for skipping tests in the current workflow by file paths changed by the pr. It is interesting enough to keep it until we find an alternative.

[pepeiborra]

So if we are keeping it, why not keep using it to cancel duplicate runs as it already does? In other words, what's the motivation for this PR?

[jneira]

So if we are keeping it, why not keep using it to cancel duplicate runs as it already does? In other words, what's the motivation for this PR?

Oh sorry, i should had linked the first pr adding it to the test workflow: #2310

In short: cancelling previous runs was not working in most cases due to permissions (when the pr is from another repo)

[jneira]

uh oh funct-tests are hanging in windows 9.0.1 here 🤦

will disable it again