This is an umbrella issue to capture some of the issues that will have to wait for the next major:

• Remove support for Node.js less than 20.19.0 and 22.13.0 #8867
• Remove deprecated CommonJS Node.js API #7395
• Remove deprecated output property in Node.js API result object #7397
• Remove deprecated GitHub formatter #8879
• Refactor to use import attributes #8880
• Change fix default to strict #8881
• Refactor to replace @dual-bundle/import-meta-resolve with import-meta-resolve #7556
• Change report to be consistent and predictable in how it handles the provided position arguments #8126
• Change selector-pseudo-*-allowed-list not to include vendor prefixes automatically #7542
• Remove resolveNestedSelectors option of selector-class-pattern #7482 (comment)

Additionally:

• Update pure ESM dependencies (wip)
• Resolve "// TODO: This test fails due to SIGSEGV on Node.js 18 for some reason."
  • Via this change in PR #8859
• Ensure that tests can run prefixed by NODE_OPTIONS='--throw-deprecation' (only 1 for context.fix)
• Unskip test disabled because of Node.js 18

Post merge:

• Remove target-branch in Dependabot config
• Add changelog entry for Bump mathml-tag-names #8874 (comment)

We'll likely want to wait at least a few months before touching them.

In the meantime, we can share any thoughts on what may be, if people have time, good things to look at in 16.x.

It'd be amazing to see:

• more rules that help people write modern CSS consistently
• a built-in resolver so that people can lint in files in context

I think rules like color-function-notation help our users write modern CSS code. I'm seeing code in the wild using modern CSS features like:

• @container
  • Add new rule to disallow specific condition in @container queries #7635
  • Fix function-no-unknown false negatives for functions in at-rule preludes #7799
• @layer
  • Add rules to specify patterns for more names #7732
• clamp() and other math comparison functions
  • function-calc-no-unspaced-operator should support newer calc functions #7618
• :has()
  • New rule: Disallow unconstrained :has() for performance reasons #7749
• logical properties & values
  • Add rules to enforce either physical or logical layout mappings  #5430
• Nesting
  • Add (*-)nesting(-*) #6421
  • Fix inaccurate end positions when postcss-resolve-nested-selector is used in selector-* #6234
  • Remove resolveNestedSelectors option of selector-class-pattern #7482
  • Fix rule-selector-property-disallowed-list false positives for nesting selectors #7488
  • Rule options where users set selectors #7557
• relative colours
  • Fix declaration-property-value-no-unknown false positives for oklch and oklab #7241 (comment)

For example:

:root {
  --dk-color-black-20: oklch(20% 0 0deg);
  --dk-color-black-20-a-10: oklch(from var(--dk-color-black-20) l c h / 10%);
}

@layer defaults {
  :where(body) {
    color: var(--dk-color-black-20);
  }

  :where(a) {
    &:any-link {
      cursor: pointer;
    }

    &:hover {
      text-decoration: none;
    }
  }
}

parent-component {
  margin-inline: auto;

  &:has(child-component) {
    container: child-component / inline-size;
    display: grid;
  }
}

child-component {
  padding: clamp(1rem, 1rem + 1cqi, 2rem);

  @container parent-component (width > 40rem) {
    display: flex;
  }
}

There are already a couple of open issues around these, and I will try to make some time to revisit them. And I'll think about what new rules we can create to help people consistently use these modern features.

The built-in resolver already has an issue (#4088). It's a biggy, though, and I've added a "needs help" label because it's likely non-trivial. It'd be great to have as it'd open up linting CSS in the context of a whole CSS codebase.

What are the things that you'd like? We can broadly discuss them here, or you can create needs: discussion issues for them.