Vale Linter

Version updated for https://github.com/vale-cli/vale-action to version v3.0.0.

  • This publisher is shown as ‘verified’ by GitHub.

  • This action is used across all versions by 4,052 repositories.

Action Type

This is a Docker action.

Go to the GitHub Marketplace to find the latest changes.

Action Summary

This GitHub Action automates the process of running Vale, a tool for static code analysis, in pull requests. It solves problems related to managing and executing Vale across various repositories by providing an official action that simplifies installation, management, and execution with ease. The key capabilities include:

  • Running Vale on Linux, macOS, and Windows
  • Handling different markup formats by installing external parsers
  • Offering suggested changes for resolved alerts
  • Using specific reporters like github-pr-review to post comments in pull requests
  • Caching style paths to speed up subsequent runs

What’s Changed

Breaking

Nothing was renamed or removed, but several behaviors changed. Each has a way back.

  • reviewdog 0.17.0 → 0.21.0. Workflows triggered on push using a github-pr-* reporter previously reported nothing (this is not PullRequest build, exit 0). They now lint, and can fail. Pin reviewdog_version: 0.17.0 to keep the old behavior.
  • The level input is honored. It was read from action.yml and never used. Checks that concluded neutral can now conclude failure. Leave level unset for the previous behavior.
  • fail_on_error: true fails on errors only. It previously failed on a finding of any severity for every reporter except the check ones, so a lone suggestion ended the run. Use fail_level: any to restore that.
  • GitHub’s 10-annotation limit no longer fails the job when fail_on_error: false. You get a warning naming how many alerts went unshown.
  • Pinning @reviewdog tracks the branch, which is now v3. Pin @v3, or @v2 to stay put.

Added

  • Suggested fixes. Alerts that Vale knows how to resolve are posted as suggested changes you can commit from the pull request — the same replacements the language server offers. Requires reporter: github-pr-review.
  • macOS, Windows, and ARM runners. The action previously only ran on Linux x86-64.
  • New inputs: config, filter, glob, min_alert_level, fail_level, sync, reviewdog_version. workdir is now declared and documented.
  • Caching. Binaries go in the runner’s tool cache, and sync: false lets you restore a cached StylesPath instead of re-downloading every package each run.
  • Downloads are verified against each release’s published checksums.

Fixed

  • Crash on large output — write EPIPE when reviewdog exited before reading it all (#122)
  • An empty files list failing with lstat : no such file or directory (#141)
  • Pull requests whose diff exceeds 20,000 lines (#153)
  • The annotation limit failing runs regardless of fail_on_error (#89, #150)
  • A warning failing the workflow under fail_on_error: true (#84)
  • Nothing reported on push events (#103)
  • The level input being ignored (#106)
  • Wildcards in files — use the glob input; the warning now says so (#123)
  • Scheduled runs on runners that write a thin event payload (#69)
  • vale_flags splitting on whitespace, which broke --filter='.Level == "error"'
  • Vale’s own error is reported when it exits with a runtime error, instead of an exit code

Internal

The action now ships as a single bundled dist/index.js; node_modules is no longer committed, taking the repository from 1,035 tracked files to 23. CI fails if the bundle is out of date with the source.

Full Changelog: https://github.com/vale-cli/vale-action/compare/2.1.2...v3.0.0