Do not try to write back ignored files after --fix.

[subzey]

Which issue, if any, is this issue related to?

Fixes issue #2627

Is there anything in the PR that needs further explanation?

Accessing opts.syntax throws if the file is ignored. I wonder why Flow didn't catch that?

[hudochenkov]

Thank you for a fix! Would you mind adding test for this issue, please? You can add it to fix.test.js.

[subzey]

@hudochenkov Looks like fix.test.js contains tests for lib version of stylelint (lib/index), and I need to test the standalone script (lib/standalone). I'm not really sure how to do that, in fact I can see no tests that contains "standalone" to start off from.

I definitely need help with tests.

[subzey]

Added a test. Not sure I did it right, but at least linters are okay, and it's green with the proposed fix code and red with the old one

[hudochenkov]

Thank you for thinking outside the box for test :)

I came up with this test:

it("doesn't write to ignored file", () => {
  return stylelint.lint({
    files: [stylesheetPath],
    config: {
      ignoreFiles: stylesheetPath,
      rules: {
        "at-rule-name-case": "lower",
        "comment-empty-line-before": "always",
        "comment-no-empty": true,
      },
    },
    fix: true,
  }).then(() => {
    return pify(fs.readFile)(stylesheetPath, "utf8").then((newFile) => {
      return pify(fs.readFile)(path.join(__dirname, "stylesheet.css"), "utf8").then((oldFile) => {
        expect(newFile).toBe(oldFile)
      })
    })
  })
})

I still don't understand promises, so this why it's ugly as hell. If someone could improve code it would be much appreciated :)

[alexander-akait]

@hudochenkov promises is very ugly if your don't use async/await 😄

[subzey]

@hudochenkov Actually, I've almost copy-pasted the code chunk from /~https://github.com/stylelint/stylelint/blob/master/system-tests/004/004.test.js#L11-L14

[jeddy3]

@hudochenkov Is this good to go in for 7.12 or did you want to change the test? Is there any advantage to either approach over the other?

[hudochenkov]

@jeddy3 it's good to merge, if test looks ok.

Is there any advantage to either approach over the other?

Test in this PR looks quite complicated. While my version might be improved. I didn't ask to use my test, because not sure if they are better than @subzey's test.

If you accept author's test, then feel free to merge.

[jeddy3]

Let's merge and then I'll drop in your test, if only because it's more consistent with the other tests in the file.

[jeddy3]

Changelog:

• Fixed: --fix no longer crashes when used with ignored files (#2652).