V9 Planned Changes

[bmish]

To release in a minor version 8.2.0 beforehand:

• feat: add type-checking with jsdoc TypeScript types and publish types #571
  • Ensure we publish types
• chore: upgrade to ESLint v9 #573

Expected major version changes:

• Enable additional recommended rules:
  • New Rule: no-invalid-names #298
• Rename recommended config to recommended-legacy, move flat config into exported configs object as recommended, remove string export for flat configs (per Change Request: Standardized export of flat configs to enable programmatic detection eslint/eslint#18095 (comment))
• Drop Node 16 support (end-of-life 2023-06-01), drop Node 19 support, use ^18.18.0 || ^20.9.0 || >=21.1.0" to match ESLint
  • We can drop additional versions, see this example: feat!: require Node 20, 22, 24+ eslint-community/eslint-plugin-eslint-plugin#529
• Any dependency upgrades

Let's ensure that both V8 and V9 are fully compatible with ESLint v8 and ESLint v9.

For reference, V8 was release 2023-06 (CHANGELOG, discussion in #222).

[jaswilli]

Hi,

I'd like to propose that require-expect be removed from the recommended set of rules. I think it's odd that we'd actually recommend against using assert.expect in tests as using it consistently makes test suites more robust and forbidding its use seems like a case of removing a set of guardrails because, "trust me, I know what I'm doing," which is sort of antithetical to the spirit of a linting tool.

I think #222 (comment) was perhaps a mischaracterization of assert-expect--I don't believe it's no longer needed nor redundant.

My actual preference would be to revert back to recommending except-simple, however, I realize that's unlikely to happen at this point 😄

[bmish]

@jaswilli that's a fair point, although I'm not yet sure I agree. Can you open up a new issue for us to discuss that?

[platinumazure]

Okay, I've gone through the open PRs and merged everything I could.

There are 3 upgrades that drop support for Node 16, so I labeled them as "breaking" for now. I think they could presumably be added to the V9 list. (I also merged an upgrade of the globals library without seeing that the library also dropped support for Node 16, but CI did pass, so... Worst case, I'll downgrade that dependency and cut a patch release if I absolutely have to.)

With all the changes done, we do have one new "feature" out since 8.0.1: A flat config version of our recommended config.

Do we think that's worth releasing in a minor release? I lean towards yes or at least maybe, but I'd like to know what others think.

Other than that, I wouldn't mind doing a major release soon just to get the Node 16 support out of the way and modernize our dependencies fully. We do have a few "help wanted" issues on the issue list, but none of those seem critical to include before or in a V9.

@bmish Let me know what you're thinking.

[bmish]

Yes definitely let's get a minor version out first.

[bmish]

We could optionally hold off on releasing v9 until ESLint v9 is out so that we can ensure that both v8 and v9 are compatible with it first.

[platinumazure]

@bmish Let's do that (since the other main compelling reason to do this, dropping support for old Node versions, is already months out of date). Thanks!

[LucasHill]

@platinumazure it should be possible to get a patch release out with eslint v9 support with this merged #532 and also should hopefully unblock your work on v9 of this plugin. Thanks!

[bmish]

We can keep the discussion about v8.1.2 in here:

• fix: eslint v9 compatibility issues #532