Don't fail on missing ~/.docker dir

[PhrozenByte]

Description

This is the then announced "might be necessary, idk…" follow-up to #39 (8828a23). See #39 for background info.

Fixes the main run() raising an ENOENT error if ~/.docker/ doesn't exist when writing the new Docker config (#42).

This time I can't really say that it happens "frequently" with GitHub-hosted runners, it has been 5 months since @divyansh42 merged my initial (and apparently incomplete) fix (#39) in March. Still no idea why ~/.docker/ sometimes doesn't exist… Still it causes workflows to fail, what still is annoying. The last error happened in a private repo's workflow, thus I can't share the link…

These changes are again UNTESTED! Simply because I still don't know how. I still know a little JavaScript, but still have next to zero experience with NodeJS (and honestly don't want to learn it), so please keep this in mind. I still believe that providing something to fix #42 is better than nothing (so many "still"s 😆).

Related Issue(s)

Fixes #42
Follow-up to #39
Related to #18
Related to #36 (maybe?)
Related to #41 (maybe?)

Checklist

• This PR includes a documentation change
• This PR does not need a documentation change

---

• This PR includes test changes
• This PR's changes are already tested
• What tests?

---

• This change is not user-facing
• This change is a patch change
• This change is a minor change
• This change is a major (breaking) change

Changes made

• Fixes run() raising an ENOENT error if ~/.docker/ doesn't exist (fixes [BUG] podman login fails when missing docker configs #42, follow-up to getDockerConfigJson(): Return empty auth when ~/.docker/config.json doesn't exist #39)

[PhrozenByte]

CI isn't running because GitHub considers this project to be too sleepy… 😆

See https://github.com/redhat-actions/podman-login/actions/workflows/ci.yml

This scheduled workflow is disabled because there hasn't been activity in this repository for at least 60 days. Enable this workflow to resume scheduled runs.

I know this very well, happened to me before, too. I've created a GitHub Action in 2022 to circumvent this, you might want to add it to this repo. There are a few other GitHub Actions in the marketplace trying to do the same, but AFAIK mine is the only that works reliably without requiring "ghost commits" or running code on some other server, but just uses GitHub workflows to keep GitHub workflows alive. See https://github.com/marketplace/actions/github-workflow-immortality

[PhrozenByte]

Looks like its getting more frequent again, I had multiple runs fail in the past few days, including a public one: (1) (as before, note that the parallel jobs doing exactly the same don't fail for some unknown reason)

Even tough I'm no fan of pinging people, but since you've merged the last one: @divyansh42, wdyt?

[mdhowle]

I have been running into this issue as well.

These changes are again UNTESTED!

I duplicated your changes in my fork. It fixed it in my case.

I still know a little JavaScript, but still have next to zero experience with NodeJS (and honestly don't want to learn it)

Amen, brother. :)