feat: Add 'ament_yamllint' package for YAML style validation

[frozenreboot]

Description

This PR adds ament_yamllint to support YAML style validation.
It revive original PR

Origianl proposal PR

#351.

Changes from original proposal

• Applied all previous review comments
• Fixed CMake export: Added CONFIG_EXTRAS to ament_package() in ament_cmake_yamllint. Without this, the ament_yamllint() macro was not visible to downstream packages.

test

1. Ran colcon test --packages-select ament_cmake_yamllint ament_yamllint (passed)
2. Integration Test: Created a local dummy package to verify behavior:
   Valid YAML files -> Pass
   Invalid/Broken YAML files -> Fail (correctly reported syntax errors and lint issues)
   Directories with COLCON_IGNORE -> Skipped

Relate issue
#412.
ros-navigation/navigation2#5853, where the lack of a standard YAML linter in ROS 2 was highlighted.

[ahcorde]

@cottsay do you mind to take a look ?

[frozenreboot]

@ahcorde Thanks for the detailed review!

I have applied all the requested changes:

• Bumped version to 0.20.3 (matching Rolling)
• Updated Copyright year to 2026
• Added author and maintainer information correctly
• Fixed setup.py metadata and dependencies as suggested

The CI checks passed successfully. It is ready for re-review.

[cottsay]

A few comments.

For the record, I pretty much just copied the original PR from another ament_lint tool and modified it for use with yamllint. I never pushed the PR forward because there wasn't really any outside interest in the tool at the time and most of Chris' feedback was applicable to the other ament_lint tool I had copied the code from.

Happy to help push this forward now that somebody's interested, though.

[frozenreboot]

I have addressed all the review comments. Ready for re-review.

[cottsay]

A couple more changes, but this is looking good. I'll trigger some CI builds on the next round of changes.

The switch for Rolling to RHEL 10 is imminent, and I'm still working on yamllint support there, so we may have to hold off a bit until that work lands.

Thanks!

[cottsay]

I believe this is a copy/paste error from another extension. The argument doesn't appear to be used and isn't documented.

Suggested change

	cmake_parse_arguments(ARG "" "MAX_LINE_LENGTH;TESTNAME" "" ${ARGN})
	cmake_parse_arguments(ARG "" "TESTNAME" "" ${ARGN})

[cottsay]

Let's order these alphabetically.

Suggested change

	LABELS "yamllint;linter"
	LABELS "linter;yamllint"

[frozenreboot]

Thanks for the review! I've addressed the feedback in the latest commit:

• Removed the unused MAX_LINE_LENGTH argument.
• Reordered test labels to be alphabetical (linter;yamllint).

I also noted your comment about the RHEL 10 transition. I'm ready to wait until the infrastructure work lands, so please take your time triggering the CI.

[cottsay]

Cross-link to ros2/ci#847

I'll trigger CI once that change merges.

[cottsay]

• Linux
• Linux-aarch64
• Linux-rhel
• Windows

[ahcorde]

I can see some tests failing on windows:

test\test_yamllint.py:22: in test_yamllint
    rc = main(argv=[])
ament_yamllint\main.py:104: in main
    report = invoke_yamllint(files, config_file=temp_config.name)
ament_yamllint\main.py:244: in invoke_yamllint
    raise subprocess.CalledProcessError(
E   subprocess.CalledProcessError: Command '['C:\\pixi_ws\\.pixi\\envs\\default\\Scripts\\yamllint.EXE', '-s', '-f', 'parsable', '-c', 'C:\\Users\\ContainerAdministrator\\AppData\\Local\\Temp\\yamllint__s5pfsoe.yaml', 'ament_yamllint\\configuration\\yamllint.yaml']' returned non-zero exit status 1.

[frozenreboot]

@ahcorde

Thanks for your review.

I noticed the Windows build failure.

The error from yamllint (returned non-zero exit status 1) suggests it might be an issue with the configuration or line endings (CRLF) on the Windows runner.

Would you be open to me adding a .gitattributes file to enforce LF for YAML files to see if that resolves it? Or is there a preferred way to handle this in this repository?

[frozenreboot]

@cottsay

I have updated the PR to fix the Windows CI failure.

The root cause was a PermissionError specific to Windows'file locking mechanism. Previously, the NamedTemporaryFile was kept open within a with block while subprocess.run tried to read it using yamllint.exe. Windows denies access to the subprocess in this state.

For context, here is the exact trace from the Windows runner:

Click to expand error log

  File "C:\hostedtoolcache\windows\Python\3.10.11\x64\lib\site-packages\yamllint\cli.py", line 197, in run
    conf = YamlLintConfig(file=args.config_file)
  File "C:\hostedtoolcache\windows\Python\3.10.11\x64\lib\site-packages\yamllint\config.py", line 41, in __init__
    with open(file, mode='rb') as f:
PermissionError: [Errno 13] Permission denied: 'C:\\Users\\RUNNER~1\\AppData\\Local\\Temp\\yamllint_fp73nw0q.yaml'

I changed the logic to explicitly close() the temporary file before invoking the subprocess, and then manually clean it up using os.remove() in a finally block. I also verified this fix on my fork using a Windows GitHub Actions runner, and it passes perfectly.

Could you please re-trigger the CI?

[frozenreboot]

Hi @cottsay,
Just a gentle ping on this PR.
The Windows PermissionError issue has been resolved in my latest commit.
Could you please re-trigger the CI when you have a chance? Thank you!