Rule Change: new ignoredLabels option for no-missing-label-refs

[lordrovo]

What rule do you want to change?

no-missing-label-refs

What change do you want to make?

Generate fewer warnings

How do you think the change should be implemented?

A new option

Example code

- [ ] Unchecked  
- [x] Checked  
- [-] In progress

What does the rule currently do for this code?

In my use case, I use GFM task lists with custom states, such as [-] to indicate "in progress".
Currently, the rule reports an error for [-]: Label reference '-' not found. This is not helpful, since [-] is not intended to be a reference label, but a visual indicator for task status.

What will the rule do after it's changed?

With the following configuration: ["error", { "ignoredLabels": ["-"] }]
The rule will ignore [-] and not report it as a missing label reference.

Participation

• I am willing to submit a pull request to implement this change.

Additional comments

No response

[snitin315]

I think that makes sense. Would like to know what the rest of the team thinks. @eslint/eslint-team

[nzakas]

@lordrovo can you try setting language: "markdown/gfm"? I'm unable to reproduce the problem.

See https://stackblitz.com/edit/stackblitz-starters-jqcrssas?file=eslint.config.js

[snitin315]

Just saw that markdownlint also has the ignored_labels option for the same use case.

But yeah, it's not reproducible with language: "markdown/gfm".

[lumirlumir]

This issue is slightly different in context, but I believe it aligns with #294.

The idea of adding an ignored_labels (or allowedLabels) option to no-missing-label-refs was also discussed in #294.

In my opinion, it would be good to discuss this issue together with #294.

[lordrovo]

@nzakas I've created a StackBlitz reproduction of this issue.

[nzakas]

@lordrovo thanks for the repro case. I see what you mean now.

I don't think ignoreLabels is the right solution to this problem because in this context it's not a label. Instead, we should just not check anything inside of a listItem node.

[JoshuaKGoldberg]

- is not intended to be a reference label, but a visual indicator for task status.

🤔 is there precedent for this anywhere? I've never seen this syntax used before, and it doesn't seem supported by GFM.

[-] on its own is a valid label ref, even inside a list item:

• -

My understanding is that the rule is actually working as intended.

[nzakas]

@JoshuaKGoldberg That's a good catch! The [-] is actually treated as a label ref if there is a reference named -.

Without:

With:

So I agree with you, the rule is behaving as expected.

And then I agree with @lumirlumir that we should align these proposal with #294 and use either allowLabels or ignoreLabels (we use the verb form "allow" and "ignore" rather than the adjective form "allowed" and "ignored" in property names).

[lordrovo]

I prefer ignoreLabels since we're specifically excluding certain patterns from being treated as label references. As mentioned above, this naming also aligns with markdownlint's ignored_labels option.