Actions: set-env depreciated

[ewels]

GitHub Actions is depreciating the set-env method previously used to pass environment variables between workflow steps.

I think that we use this method in quite a few places, so we may need to make some updates.

See https://github.blog/changelog/2020-10-01-github-actions-deprecating-set-env-and-add-path-commands/

[ewels]

This is going away November 16, so is now urgent: https://github.blog/changelog/2020-11-09-github-actions-removing-set-env-and-add-path-commands-on-november-16/

[ggabernet]

nf-core tools

[ewels]

The add-path command is deprecated and will be disabled on November 16th. Please upgrade to using Environment Files. For more information see: https://github.blog/changelog/2020-10-01-github-actions-deprecating-set-env-and-add-path-commands/

https://github.com/nf-core/rnaseq/actions/runs/360361285

[ggabernet]

Maybe that's something we can test really nicely in the testpipeline.

In the case of the awsmegatests, it would hopefully be a matter only of updating the setup miniconda action to conda-incubator/setup-miniconda@v2 here.

[ggabernet]

And the CI workflow uses the set-env command, not sure if part of the workflow or of the imported actions (e.g. technote-space/get-diff-action@v1): https://github.com/nf-core/testpipeline/actions/runs/357450322

[KevinMenden]

Using conda-incubator as suggested by Gisela did the trick for the awstest and awsmegatest files.

Using a newer version of get-diff-action should solve the trick for ci.yml as well. Needed to change the PREFIX_FILTER command with FILES due to the newer version, too.
I'm getting some new errors new when testing on the testpipeline but they seem to be docker related ....
https://github.com/nf-core/testpipeline/actions/runs/361878129

[ewels]

Looks like the linting is failing: nf-core/testpipeline#12 (comment)

❌ Test failures:

• Test #5 - GitHub Actions 'branch' workflow should be triggered for PRs to master: /home/runner/work/testpipeline/testpipeline/.github/workflows/branch.yml

If the linting fails then other CI checks may also be stopped early..

Not entirely sure what the failure is due to. Could be a remnant from me mucking around with stuff earlier? Might be worth rolling back my commits so that you're working with an otherwise fresh repo again.

[KevinMenden]

The CI failure is due to the docker container, which it tries to build. I could replicate it on my machine:

ERROR conda.core.link:_execute(502): An error occurred while installing package 'conda-forge::click-7.1.2-pyh9f0ad1d_0'.
OSError(2, 'No such file or directory')
Attempting to roll back.

It's probably something with the order of installation in bioconda again :/

[KevinMenden]

Okay so the error comes when installing multiqc in the docker container ... which is odd because that should be working fine ...

[KevinMenden]

The issue was the base image ... when using the nfcore/dev base image, it fails. Will test again and if CI works I can make a PR to solve the 'set-env' issue here.

[ewels]

Just had my first CI error due to this being disabled: https://github.com/nf-core/smrnaseq/pull/57/checks?check_run_id=1408805300 😱

Will try updating that pipeline to have your changes and hope it fixes it!

[ewels]

Nice! https://github.com/nf-core/smrnaseq/pull/57/checks?check_run_id=1408826335 🚀

Great work @KevinMenden!

[ewels]

Solved, just needs a release.