trained-models Replace git checkout with datalad checkout in action workflows wherever appropriate

Replace git checkout with datalad checkout in action workflows wherever appropriate

Open hvgazula opened this issue 1 year ago • 6 comments

It makes much more sense and is cleaner to use datalad checkout instead of git checkout as the siblings are not readily known with the latter.

Feb 13 '24 15:02 hvgazula

WIP https://github.com/datalad/datalad-action/issues/18#issuecomment-1942666487

Feb 13 '24 22:02 hvgazula

Works alright in GitHub runner but not on ec2 self-hosted runner because all actions are run as root.

Feb 14 '24 07:02 hvgazula

changing no non-root user didn't cut either. https://github.com/neuronets/test-aws/actions/runs/7904098846

Feb 14 '24 16:02 hvgazula

I am giving up. Thanks.

Feb 14 '24 17:02 hvgazula

why not just log into an ec2 instance and carry out the steps manually?

Feb 14 '24 19:02 satra

Nah..I am testing the action itself not simulating the steps in the action...

Feb 14 '24 20:02 hvgazula