chore: update minimatch

[Pixel998]

Prerequisites checklist

• I have read the contributing guidelines.

What is the purpose of this pull request?

This PR updates the minimatch package to the latest version

What changes did you make? (Give an overview)

• Updated minimatch to v10.0.3
• Switched to named imports for minimatch, in line with the updated API.
• Removed the deprecated @types/minimatch package since minimatch now includes its own built-in TypeScript type definitions.

minimatch v10 requires Node.js v20 or higher, so marking this PR as draft until we drop Node.js v18.

Related Issues

Fixes #230
Fixes #66

Is there anything you'd like reviewers to focus on?

[mdjermanovic]

Can we now remove allowWindowsEscape: true, option, since that behavior should be on by default in minimatch v10?

rewrite/packages/config-array/src/config-array.js

Line 65 in 88b0c34

allowWindowsEscape: true,

[mdjermanovic]

We should also update minimatch dependency in the eslint package:

https://github.com/eslint/eslint/blob/e1ac05e2fae779e738f85bd47dda1cc2b7099346/package.json#L139

[mdjermanovic]

Are there any other breaking changes we should be concerned about in these 7 major versions?

[Pixel998]

I think there are two main breaking changes we need to be aware of beyond the obvious Node.js version bumps and export changes:

1- POSIX character class support (added in v7.3)

Example test that will fail in 3.x but pass in 10.x:

describe("POSIX character classes", () => {
  it("[[:alpha:]] matches alphabetic characters", () => {
    const configs = new ConfigArray(
      [
        {
          files: ["**/[[:alpha:]].js"],
        },
      ],
      { basePath },
    );

    configs.normalizeSync();

    assert.strictEqual(configs.getConfigStatus("a.js"), "matched");
  });
});

2- Pattern simplification with .. segments

Minimatch added an optimizationLevel option that controls how patterns are normalized before matching:

• 0 → No changes. . and .. must appear literally in both the pattern and the test string.
  Example: a/*/../c matches a/b/../c, but not a/c.
• 1 (default) → Simplify .. segments when possible.
  Example: ./a/b/../* is simplified to ./a/*, so it matches ./a/c, but not ./a/b/../c.

Example test that will fail in 3.x but pass in 10.x:

describe("Pattern simplification with '..' segments", () => {
  it("'a/b/../*.js' behaves like 'a/*.js'", () => {
    const configs = new ConfigArray(
      [
        {
          files: ["a/b/../*.js"],
        },
      ],
      { basePath },
    );

    configs.normalizeSync();

    assert.strictEqual(configs.getConfigStatus("a/x.js"), "matched");
    assert.strictEqual(configs.getConfigStatus("a/b/x.js"), "unconfigured");
  });

  it("'a/b/../**/*.js' behaves like 'a/**/*.js'", () => {
    const configs = new ConfigArray(
      [
        {
          files: ["a/b/../**/*.js"],
        },
      ],
      { basePath },
    );

    configs.normalizeSync();

    assert.strictEqual(configs.getConfigStatus("a/x/y.js"), "matched");
  });
});

[mdjermanovic]

Thanks for the detailed analysis!

Example test that will fail in 3.x but pass in 10.x:

describe("POSIX character classes", () => {
  it("[[:alpha:]] matches alphabetic characters", () => {
    const configs = new ConfigArray(
      [
        {
          files: ["**/[[:alpha:]].js"],
        },
      ],
      { basePath },
    );

    configs.normalizeSync();

    assert.strictEqual(configs.getConfigStatus("a.js"), "matched");
  });
});

Since this is a new feature that will become available in config-array and consequently in ESLint, but at the same time kind of breaking because [[:alpha:]].js can be a valid filename in posix (I think?), it might be best to tag this PR as feat!?

[Pixel998]

We should also update minimatch dependency in the eslint package

Yeah, it’s on my todo list — I’ll get to it today or tomorrow.

[aladdin-add]

Yeah, it’s on my todo list — I’ll get to it today or tomorrow.

@Pixel998 I can take it(if you don't have time to look at it.)🙌