I saw that the tests/* directory wasn't being ran through shellcheck whatsoever, so thanks for getting to this before I was able to.

It looks like /lib/asdf.sh is sourced by /asdf.sh so I'd say the file extension shouldn't be Bash.

EDIT: Also for the files that end in .sh, there should be a shellcheck declaration like # shellcheck shell=sh on the first line so it can lint properly (in the editor)

I have certainly missed things, still correcting and updating. This is a rabbit hole uncovering a lot of issues 😱 Haven't even started on resolving the issues.

EDIT:

It looks like /lib/asdf.sh is sourced by /asdf.sh so I'd say the file extension shouldn't be Bash.

Both asdf.sh & lib/asdf.sh throw errors when running Shellcheck with sh, because they're using Bash features. So should we change their names, or remove the Bash from the code, or do nothing?

Both asdf.sh & lib/asdf.sh throw errors when running Shellcheck with sh, because they're using Bash features. So should we change their names, or remove the Bash from the code, or do nothing?

I'd say we keep their names and keep the declaration at the top. Since we want CI to pass for everyone, we can just add # shellcheck disable=... lines before each Bashism for now, until we decide what to do with them later.

until we decide what to do with them later

Right, I was wondering if we should just solve that now. Happy to ignore specifics for these files now.

Oh wow yeah, there are 5 real violations in those files, yeah. In my opinion most of the file should be reworked anyways. The variables ASDF_BIN ASDF_USER_SHIMS shouldn't be capitalized and not all the variables are prefixed with an underscore. And I'm not sure why we're munging PATH - if really we don't want to append to the path twice, we can just set an environment variable with values (but a naive approach will break in several circumstances). So I guess that's to say I'm for doing everything separately, since there wil be more eyes and especially since if there is one mistake in this crutial code path, everything likely could break

Yes, there are a lot of errors here.

@Stratus3D My suggested path forward is this:

• merge fix: lint errors from scripts/checkstyle.py #1385
• rebase this PR
• disable the check of asdf.sh & lib/asdf.sh
• go through and fix all reported issues (I can do this)
• merge this PR
• rebase all other PRs
• unlock and tackle asdf.sh & lib/asdf.sh in a separate PR

What do you think?

I have commented out the updates to scripts/* that cause new errors, this way we can introduce them piecemeal over the next month or so. This way the branch won't have sweeping changes and become stale and reduce disruption to other PRs.

I will create issues to track the uncommenting of these scripts.