Curtis Vogt

Have you validated this is working? This failed CI run didn't seem to save the cache: - Failed job: https://github.com/julia-actions/cache/actions/runs/7572229335/job/20667038487?pr=107#step:11:123 - Successful job: https://github.com/julia-actions/cache/actions/runs/7572229335/job/20666823090?pr=107#step:13:211

Based upon [this comment](https://github.com/julia-actions/cache/pull/123#pullrequestreview-1981421906) the change to `actions/cache@v4` should be considered breaking here. I think we'll be merging #123 first which means this change will be non-breaking for the upcoming...

The quick work around for this is to run `Pkg.Registry.update()` immediately after `julia-actions/cache` to ensure the registries are up to date before running `julia-actions/runtest`.

Usually the `post` step is conditionally run based upon the [`post-if`](https://docs.github.com/en/actions/creating-actions/metadata-syntax-for-github-actions#runspost-if) setting. As [`pyTooling/Actions/with-post-step` doesn't specify this](https://github.com/pyTooling/Actions/blob/60281e01e2514114849e490153aa0a3627994077/with-post-step/action.yml#L42) I'm not sure we can fix can this beyond setting `save-always`.

I think the way we'll have to address this is to always leave the latest cache entry remaining. That way if we are cleaning up cache entries on a failed...

> If you don't the cache won't be saved in re-runs. Aren't we just saving redundant information though? What can change between re-runs such that we need to save a...

Alternatively, if we're willing to switch away from using a composite action we can use [`runs.post`](https://docs.github.com/en/actions/creating-actions/metadata-syntax-for-github-actions#runspost) to perform the cache save (unfortunately, this is basically re-implementing `actions/cache`). It's rather unfortunate...

It's out in the `v4` release: https://github.com/actions/cache#v4

Another example: ``` Invalid instruction at 0x75fc0a81d157: 0x62, 0xf2, 0x7d, 0x48, 0x7c, 0xc0, 0x62, 0xf1, 0x7d, 0x48, 0xfe, 0x0d, 0x99, 0x40, 0xfe [1732] signal (4.2): Illegal instruction ```

I ended up displaying `/proc/cpuinfo` in my workflow and found that using GitHub hosted runners for `ubuntu-latest` do indeed switch between Intel and AMD CPUs. In my particular case runs...