Repository navigation
Code spans and fenced blocks in plain comments are linted as prose #1208
Description
Check for existing issues
- Completed
Environment
macOS (arm64), built from source, v3 at cfb21a4.
Describe the bug / provide steps to reproduce it
In a comment read as plain lines, a word is only treated as code when it touches a backtick. A word in the middle of a code span is linted as prose, and so is everything in a fenced example. With an existence rule on ZEBRA and this test.ts:
/**
* Reads it.
*
* ```ts
* const ZEBRA = read();
* ```
*/
export function f() {}
// A host (`fr_FR` → `jeu. 30 ZEBRA …`) fails, and ZEBRA is prose.
f();vale test.ts reports 5:10 (inside the fence) and 10:31 (inside the span) as well as 10:52; only 10:52 is prose.
I have a fix with tests at v3...chad-loder:vale:fix/comment-code-masks. It masks these the way the doc-comment conventions already mask tags, so it depends on the fix for #1207. I'm holding both PRs while #1204 is open, per CONTRIBUTING: "Finish one before opening the next."
Activity
jdkato commentedon Oct 16, 2026
A comment with no format assigned is linted as plain lines, and in plain text a backtick is a character. Code spans and fences are Markdown; assign it and they're handled:
chad-loder commentedon Oct 16, 2026
Thank you, that works. With
ts = mdthe repro reports only10:52, so my fix branch isn't needed.For anyone who finds this later: the comment is then read as Markdown throughout, so a line indented four or more spaces after a blank line is an indented code block and is skipped, just as it would be in a
.mdfile. Most doc comments are already written that way, so it rarely comes up.