Reading Notes #658

This week, we explore the latest insights on AI, Cloud, and software development to keep you informed and inspired.

little branch with leaves and walnuts

Cloud

Programming

Databases

AI

Miscellaneous


Sharing my Reading Notes is a habit I started a long time ago, where I share a list of all the articles, blog posts, and books that catch my interest during the week. 

If you have interesting content, share it! 

~frank

Why Your .NET Code Coverage Badge is 'Unknown' in GitLab (And How to Fix It)


In a recent post, I shared how to set up a CI/CD pipeline for a .NET Aspire project on GitLab. The pipeline includes unit tests, security scanning, and secret detection, and if any of those fail, the pipeline would fail. Great, but what about code coverage for the unit tests? The pipeline included code coverage commands, but the coverage was not visible in the GitLab interface. Let's fix that.

(blog post en français ici)

Badge on Gitlab showing coverage unknown

The Problem

One thing I initially thought was that the regex used to extract the coverage was incorrect. The regex used in the pipeline was:

coverage: '/Total\s*\|\s*(\d+(?:\.\d+)?)%/'

That regex came directly from the GitLab documentation, so I thought it should work correctly. However, coverage still wasn't visible in the GitLab interface.

So with the help of GitHub Copilot, we wrote a few commands to validate:

  • That the coverage.cobertura.xml was in a consistent location (instead of being in a folder with a GUID name)
  • That the coverage.cobertura.xml file was in a valid format
  • What exactly the regex was looking for

Everything checked out fine, so why was the coverage not visible?

The Solution

It turns out that the coverage command with the regex expression is scanning the console output and not the coverage.cobertura.xml file. Aha! One solution was to install dotnet-tools to changing where the the test results was persisted; to the console instead of the XML file, but I preferred keeping the .NET environment unchanged.

The solution I ended up implementing was executing a grep command to extract the coverage from the coverage.cobertura.xml file and then echoing it to the console. Here's what it looks like:

- COVERAGE=$(grep -o 'line-rate="[0-9.]*"' TestResults/coverage.cobertura.xml | head -1 | grep -o '[0-9.]*' | awk '{printf "%.1f", $1*100}')
- echo "Total | ${COVERAGE}%"

Results

And now when the pipeline runs, the coverage is visible in the GitLab pipeline!



And the badge is updated to show the coverage percentage.

Coverage badge showing percentage


Complete Configuration

Here's the complete test job configuration. Of course, the full .gitlab-ci.yml file is available in the GitLab repository.

test:
  stage: test
  image: mcr.microsoft.com/dotnet/sdk:9.0
  <<: *dotnet_cache
  dependencies:
    - build
  script:
    - dotnet test $SOLUTION_FILE --configuration Release --logger "junit;LogFilePath=$CI_PROJECT_DIR/TestResults/test-results.xml" --logger "console;verbosity=detailed" --collect:"XPlat Code Coverage" --results-directory $CI_PROJECT_DIR/TestResults
    - find TestResults -name "coverage.cobertura.xml" -exec cp {} TestResults/coverage.cobertura.xml \;
    - COVERAGE=$(grep -o 'line-rate="[0-9.]*"' TestResults/coverage.cobertura.xml | head -1 | grep -o '[0-9.]*' | awk '{printf "%.1f", $1*100}')
    - echo "Total | ${COVERAGE}%"
  artifacts:
    when: always
    reports:
      junit: "TestResults/test-results.xml"
      coverage_report:
        coverage_format: cobertura
        path: "TestResults/coverage.cobertura.xml"
    paths:
      - TestResults/
    expire_in: 1 week
  coverage: '/Total\s*\|\s*(\d+(?:\.\d+)?)%/'

Conclusion

I hope this helps others save time when setting up code coverage for their .NET projects on GitLab. The key insight is that GitLab's coverage regex works on console output, not on the files (XML or other formats).

If you have any questions or suggestions, feel free to reach out!


~frank



Reading Notes #657

a rocky path ends at the shore of a lake
This week's collection of interesting articles and resources covers AI development, DevOps practices, and open source tools. From GitHub Copilot customization to local AI deployments and containerization best practices, here are the highlights worth your attention.

AI

DevOps

  • Local Deploy with Bicep (Sam Cogan) - A perfect short story, I'll explain why the hell bicep can now deploy locally and how to do it

Open Source

  • Introducing OpenCLI (Patrik Svensson) - A standard that describes CLI so both humans and agents can understand how it works. Love it!

~frank