Use eager Pkg server registry instead of git cloning

[fredrikekre]

Simpler alternative to #29 and #30. If one wants the conservative registry one can configure that by setting the environment variable globally intead. Will need a companion PR to julia-actions/runtests (julia-actions/julia-runtest#74)

[codecov]

Codecov Report

Patch and project coverage have no change.

Comparison is base (252d468) 100.00% compared to head (46deeb4) 100.00%.

Additional details and impacted files

@@            Coverage Diff            @@
##            master       #31   +/-   ##
=========================================
  Coverage   100.00%   100.00%           
=========================================
  Files            1         1           
  Lines            2         2           
=========================================
  Hits             2         2           

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

☔ View full report in Codecov by Sentry.
📢 Do you have feedback about the report comment? Let us know in this issue.

[SaschaMann]

Thanks! This does seem a lot simpler than the alternative PRs.

I don't know how the Pkg servers/registries work well enough to review this properly but the GHA parts look fine.

[DilumAluthge]

This is not sufficient, which is why I was working on #30.

With this PR, here is what will happen. Suppose that the user does not set the JULIA_PKG_SERVER_REGISTRY_PREFERENCE environment variable in their GitHub workflow file.

1. This action runs, which gives the user the eager reference.
2. In a subsequent step, the user calls e.g. Pkg.add(), or some other registry operation that updates the registry. Because the JULIA_PKG_SERVER_REGISTRY_PREFERENCE environment variable is not set, Pkg will use the conservative registry. If the conservative registry is behind the eager registry, this means that Pkg will overwrite the eager registry with the conservative registry.

The point is, if you want to use the eager registry for a CI job, it's not sufficient to just set the environment variable for one step in the job. You have to make sure that the environment variable is set for ALL future steps in the job, because if a future step runs a Pkg operation, and the environment variable is not set, that Pkg operation will restore the conservative registry (overwriting the previous eager registry).

[fredrikekre]

True. julia-actions/julia-runtest#74 does it for runtest so for the vast majority of packages things are OK already.

In any case, it doesn't hurt to export the env var unconditionally actually.

[DilumAluthge]

In any case, it doesn't hurt to export the env var unconditionally actually.

Yes, which is what I was working on in #30 (exporting the environment variable by default, but having an option for the user to disable the environment variable propogation).

[SaschaMann]

but having an option for the user to disable the environment variable propogation

We don't need an option for that, I think. They should be able to overwrite it by setting the env var in their workflows if for whatever reason they want to switch between the eager/conservative registry within the same workflow. Otherwise export whatever is being used here.

[fredrikekre]

Agreed, it is better to tell the CI config what to do instead of what to not do. I will submit a PR.

[fredrikekre]

#33 + julia-actions/julia-runtest#75