Documentation testing

GitLab documentation is stored in projects with code and treated like code. Therefore, we use processes similar to those used for code to maintain standards and quality of documentation.

We have tests:

  • To lint the words and structure of the documentation.
  • To check the validity of internal links within the documentation suite.
  • To check the validity of links from UI elements, such as files in app/views files.

For the specifics of each test run in our CI/CD pipelines, see the configuration for those tests in the relevant projects:

We also run some documentation tests in the GitLab Development Kit project: https://gitlab.com/gitlab-org/gitlab-development-kit/-/blob/main/.gitlab/ci/test.gitlab-ci.yml.

Run tests locally

Similar to previewing your changes locally, you can also run these tests on your local computer. This has the advantage of:

  • Speeding up the feedback loop. You can know of any problems with the changes in your branch without waiting for a CI/CD pipeline to run.
  • Lowering costs. Running tests locally is cheaper than running tests on the cloud infrastructure GitLab uses.

To run tests locally, it’s important to:

Lint checks

Lint checks are performed by the lint-doc.sh script and can be executed as follows:

  1. Navigate to the gitlab directory.
  2. Run:

    MD_DOC_PATH=path/to/my_doc.md scripts/lint-doc.sh
    

Where MD_DOC_PATH points to the file or directory you would like to run lint checks for. If you omit it completely, it defaults to the doc/ directory. The output should be similar to:

=> Linting documents at path /path/to/gitlab as <user>...
=> Checking for cURL short options...
=> Checking for CHANGELOG.md duplicate entries...
=> Checking /path/to/gitlab/doc for executable permissions...
=> Checking for new README.md files...
=> Linting markdown style...
=> Linting prose...
✔ 0 errors, 0 warnings and 0 suggestions in 1 file.
✔ Linting passed

This requires you to either:

  • Have the required lint tools installed on your computer.
  • A working Docker installation, in which case an image with these tools pre-installed is used.

Merge requests containing changes to Markdown (.md) files run a docs-lint links job, which runs two types of link checks. In both cases, links with destinations that begin with http or https are considered external links, and skipped:

  • bundle exec nanoc check internal_links: Tests links to internal pages.
  • bundle exec nanoc check internal_anchors: Tests links to subheadings (anchors) on internal pages.

Failures from these tests are displayed at the end of the test results in the Issues found! area. For example, failures in the internal_anchors test follow this format:

[ ERROR ] internal_anchors - Broken anchor detected!
  - source file `/tmp/gitlab-docs/public/ee/user/application_security/api_fuzzing/index.html`
  - destination `/tmp/gitlab-docs/public/ee/development/code_review.html`
  - link `../../../development/code_review.html#review-response-slo`
  - anchor `#review-response-slo`
  • Source file: The full path to the file containing the error. To find the file in the gitlab repository, replace /tmp/gitlab-docs/public/ee with doc, and .html with .md.
  • Destination: The full path to the file not found by the test. To find the file in the gitlab repository, replace /tmp/gitlab-docs/public/ee with doc, and .html with .md.
  • Link: The actual link the script attempted to find.
  • Anchor: If present, the subheading (anchor) the script attempted to find.

Check for multiple instances of the same broken link on each page reporting an error. Even if a specific broken link appears multiple times on a page, the test reports it only once.

To execute documentation link tests locally:

  1. Navigate to the gitlab-docs directory.
  2. Run the following commands:

    # Check for broken internal links
    bundle exec nanoc check internal_links
    
    # Check for broken external links (might take a lot of time to complete).
    # This test is set to be allowed to fail and is run only in the gitlab-docs project CI
    bundle exec nanoc check internal_anchors
    

The ui-docs-links lint job uses haml-lint to test that all documentation links from UI elements (app/views files, for example) are linking to valid pages and anchors.

To run the ui-docs-links test locally:

  1. Open the gitlab directory in a terminal window.
  2. Run:

    bundle exec haml-lint -i DocumentationLinks
    

If you receive an error the first time you run this test, run bundle install, which installs the dependencies for GitLab, and try again.

If you don’t want to install all of the dependencies to test the links, you can:

  1. Open the gitlab directory in a terminal window.
  2. Install haml-lint:

    gem install haml_lint
    
  3. Run:

    haml-lint -i DocumentationLinks
    

If you manually install haml-lint with this process, it does not update automatically and you should make sure your version matches the version used by GitLab.

Update linter configuration

Vale configuration and markdownlint configuration is under source control in each project, so updates must be committed to each project individually.

We consider the configuration in the gitlab project as the source of truth and that’s where all updates should first be made.

On a regular basis, the changes made in gitlab project to the Vale and markdownlint configuration should be synchronized to the other projects. In omnibus-gitlab, gitlab-runner, and charts/gitlab:

  1. Create a new branch.
  2. Copy the configuration files from the gitlab project into this branch, overwriting the project’s old configuration. Make sure no project-specific changes from the gitlab project are included. For example, RelativeLinks.yml is hard coded for specific projects.
  3. Create a merge request and submit it to a technical writer for review and merge.

Update linting images

Lint tests run in CI/CD pipelines using images from the gitlab-docs container registry.

If a new version of a dependency is released (like a new version of Ruby), we should update the images to use the newer version. Then, we can update the configuration files in each of our documentation projects to point to the new image.

To update the linting images:

  1. In gitlab-docs, open a merge request to update .gitlab-ci.yml to use the new tooling version. (Example MR)
  2. When merged, start a Build docs.gitlab.com every hour scheduled pipeline.
  3. Go the pipeline you started, and manually run the relevant build-images job, for example, image:docs-lint-markdown.
  4. In the job output, get the name of the new image. (Example job output)
  5. Verify that the new image was added to the container registry.
  6. Open merge requests to update each of these configuration files to point to the new image. In each merge request, include a small doc update to trigger the job that uses the image.
  7. In each merge request, check the relevant job output to confirm the updated image was used for the test. (Example job output)
  8. Assign the merge requests to any technical writer to review and merge.

Local linters

To help adhere to the documentation style guidelines, and improve the content added to documentation, install documentation linters and integrate them with your code editor.

At GitLab, we mostly use:

markdownlint

markdownlint checks that Markdown syntax follows certain rules, and is used by the docs-lint test.

Our Documentation Style Guide and Markdown Guide elaborate on which choices must be made when selecting Markdown syntax for GitLab documentation. This tool helps catch deviations from those guidelines.

markdownlint configuration is found in the following projects:

This configuration is also used in build pipelines.

You can use markdownlint:

Vale

Vale is a grammar, style, and word usage linter for the English language. Vale’s configuration is stored in the .vale.ini file located in the root directory of projects.

Vale supports creating custom tests that extend any of several types of checks, which we store in the .linting/vale/styles/gitlab directory in the documentation directory of projects.

You can find Vale configuration in the following projects:

This configuration is also used in build pipelines, where error-level rules are enforced.

You can use Vale:

Vale result types

Vale returns three types of results:

  • Error - For branding guidelines, trademark guidelines, and anything that causes content on the docs site to render incorrectly.
  • Warning - For Technical Writing team style preferences.
  • Suggestion - For basic technical writing tenets and best practices.

The result types have these attributes:

Result typeDisplayed in CI/CD job outputCauses CI/CD jobs to failVale rule link
error Yes YesError-level Vale rules
warning No NoWarning-level Vale rules
suggestion No NoSuggestion-level Vale rules

Vale spelling test

When Vale flags a valid word as a spelling mistake, you can fix it following these guidelines:

Flagged wordGuideline
jargonRewrite the sentence to avoid it.
correctly-capitalized name of a product or serviceAdd the word to the vale spelling exceptions list.
name of a personRemove the name if it’s not needed, or add the vale exception code in-line.
a command, variable, code, or similarPut it in backticks or a code block. For example: The git clone command can be used with the CI_COMMIT_BRANCH variable. -> The `git clone` command can be used with the `CI_COMMIT_BRANCH` variable.
UI text from GitLabVerify it correctly matches the UI, then: If it does not match the UI, update it. If it matches the UI, but the UI seems incorrect, create an issue to see if the UI needs to be fixed. If it matches the UI and seems correct, add it to the vale spelling exceptions list.
UI text from a third-party productRewrite the sentence to avoid it, or add the vale exception code in-line.

Vale uppercase (acronym) test

The Uppercase.yml test checks for incorrect usage of words in all capitals. For example, avoid usage like This is NOT important.

If the word must be in all capitals, follow these guidelines:

Flagged wordGuideline
Acronym (likely known by the average visitor to that page)Add the acronym to the list of words and acronyms in Uppercase.yml.
Acronym (likely not known by the average visitor to that page)The first time the acronym is used, write it out fully followed by the acronym in parentheses. In later uses, use just the acronym by itself. For example: This feature uses the File Transfer Protocol (FTP). FTP is....
Correctly capitalized name of a product or serviceAdd the name to the list of words and acronyms in Uppercase.yml.
Command, variable, code, or similarPut it in backticks or a code block. For example: Use `FALSE` as the variable value.
UI text from a third-party productRewrite the sentence to avoid it, or add the vale exception code in-line.

Vale readability score

In ReadingLevel.yml, we have implemented the Flesch-Kincaid grade level test to determine the readability of our documentation.

As a general guideline, the lower the score, the more readable the documentation. For example, a page that scores 12 before a set of changes, and 9 after, indicates an iterative improvement to readability. The score is not an exact science, but is meant to help indicate the general complexity level of the page.

The readability score is calculated based on the number of words per sentence, and the number of syllables per word. For more information, see the Vale documentation.

When to add a new Vale rule

It’s tempting to add a Vale rule for every style guide rule. However, we should be mindful of the effort to create and enforce a Vale rule, and the noise it creates.

In general, follow these guidelines:

  • If you add an error-level Vale rule, you must fix the existing occurrences of the issue in the documentation before you can add the rule.

    If there are too many issues to fix in a single merge request, add the rule at a warning level. Then, fix the existing issues in follow-up merge requests. When the issues are fixed, promote the rule to an error.

  • If you add a warning-level or suggestion-level rule, consider:

    • How many more warnings or suggestions it creates in the Vale output. If the number of additional warnings is significant, the rule might be too broad.

    • How often an author might ignore it because it’s acceptable in the context. If the rule is too subjective, it cannot be adequately enforced and creates unnecessary additional warnings.

Install linters

At a minimum, install markdownlint and Vale to match the checks run in build pipelines:

  1. Install markdownlint-cli:

    yarn global add markdownlint-cli
    

    We recommend installing the version of markdownlint-cli used (see variables: section) when building the image:docs-lint-markdown.

  2. Install vale. To install for:

    • macOS using brew, run: brew install vale.
    • Linux, use your distribution’s package manager or a released binary.

These tools can be integrated with your code editor.

Update linters

It’s important to use linter versions that are the same or newer than those run in CI/CD. This provides access to new features and possible bug fixes.

To match the versions of markdownlint-cli and vale used in the GitLab projects, refer to the versions used (see variables: section) when building the image:docs-lint-markdown Docker image containing these tools for CI/CD.

ToolVersionCommandAdditional information
markdownlint-cliLatestyarn global add markdownlint-cliNone.
markdownlint-cliSpecificyarn global add markdownlint-cli@0.23.2The @ indicates a specific version, and this example updates the tool to version 0.23.2.
ValeLatestbrew update && brew upgrade valeThis command is for macOS only.
ValeSpecificNot applicable.Binaries can be directly downloaded.

Configure editors

Using linters in your editor is more convenient than having to run the commands from the command line.

To configure markdownlint in your editor, install one of the following as appropriate:

To configure Vale in your editor, install one of the following as appropriate:

  • Sublime Text SublimeLinter-contrib-vale package.
  • Visual Studio Code errata-ai.vale-server extension. You can configure the plugin to display only a subset of alerts.
  • Atom atomic-vale package.
  • Vim ALE plugin.
  • JetBrains IDEs - No plugin exists, but this issue comment contains tips for configuring an external tool.
  • Emacs Flycheck extension. This requires some configuration:

    • Flycheck supports markdownlint-cli out of the box, but you must point it to the .markdownlint.yml at the base of the project directory. A .dir-locals.el file can accomplish this:
    ;; Place this code in a file called `.dir-locals.el` at the root of the gitlab project.
    ((markdown-mode . ((flycheck-markdown-markdownlint-cli-config . ".markdownlint.yml"))))
    
    
    • A minimal configuration for Flycheck to work with Vale could look like this:
    (flycheck-define-checker vale
      "A checker for prose"
      :command ("vale" "--output" "line" "--no-wrap"
                source)
      :standard-input nil
      :error-patterns
        ((error line-start (file-name) ":" line ":" column ":" (id (one-or-more (not (any ":")))) ":" (message)   line-end))
      :modes (markdown-mode org-mode text-mode)
      :next-checkers ((t . markdown-markdownlint-cli))
    )
    
    (add-to-list 'flycheck-checkers 'vale)
    

    In this setup the markdownlint checker is set as a “next” checker from the defined vale checker. Enabling this custom Vale checker provides error linting from both Vale and markdownlint.

Configure pre-push hooks

Git pre-push hooks allow Git users to:

  • Run tests or other processes before pushing a branch.
  • Avoid pushing a branch if failures occur with these tests.

lefthook is a Git hooks manager, making configuring, installing, and removing Git hooks easy.

Configuration for lefthook is available in the lefthook.yml file for the gitlab project.

To set up lefthook for documentation linting, see Pre-push static analysis.

Show Vale warnings on push

By default, lefthook shows only Vale errors when pushing changes to a branch. The default branches have no Vale errors, so any errors listed here are introduced by commits to the branch.

To also see the Vale warnings when pushing to a branch, set a local environment variable: VALE_WARNINGS=true.

Enable Vale warnings on push to improve the documentation suite by:

  • Detecting warnings you might be introducing with your commits.
  • Identifying warnings that already exist in the page, which you can resolve to reduce technical debt.

These warnings:

  • Don’t stop the push from working.
  • Don’t result in a broken pipeline.
  • Include all warnings for a file, not just warnings that are introduced by the commits.

To enable Vale warnings on push:

  • Automatically, add VALE_WARNINGS=true to your shell configuration.
  • Manually, prepend VALE_WARNINGS=true to invocations of lefthook. For example:

    VALE_WARNINGS=true bundle exec lefthook run pre-push
    

You can also configure your editor to show Vale warnings.

Show subset of Vale alerts

You can set Visual Studio Code to display only a subset of Vale alerts when viewing files:

  1. Go to Preferences > Settings > Extensions > Vale.
  2. In Vale CLI: Min Alert Level, select the minimum alert level you want displayed in files.

To display only a subset of Vale alerts when running Vale from the command line, use the --minAlertLevel flag, which accepts error, warning, or suggestion. Combine it with --config to point to the configuration file within the project, if needed:

vale --config .vale.ini --minAlertLevel error doc/**/*.md

Omit the flag to display all alerts, including suggestion level alerts.

Disable Vale tests

You can disable a specific Vale linting rule or all Vale linting rules for any portion of a document:

  • To disable a specific rule, add a <!-- vale gitlab.rulename = NO --> tag before the text, and a <!-- vale gitlab.rulename = YES --> tag after the text, replacing rulename with the filename of a test in the GitLab styles directory.
  • To disable all Vale linting rules, add a <!-- vale off --> tag before the text, and a <!-- vale on --> tag after the text.

Whenever possible, exclude only the problematic rule and lines.

For more information, see Vale’s documentation.

Disable markdownlint tests

To disable all markdownlint rules, add a <!-- markdownlint-disable --> tag before the text, and a <!-- markdownlint-enable --> tag after the text.

To disable only a specific rule, add the rule number to the tag, for example <!-- markdownlint-disable MD044 --> and <!-- markdownlint-enable MD044 -->.

Whenever possible, exclude only the problematic lines.