Skip to content

CONTRIBUTING.md

Latest commit

 

History

History
43 lines (28 loc) · 2.24 KB

CONTRIBUTING.md

File metadata and controls

43 lines (28 loc) · 2.24 KB
Raw

Contributing

Contributions are welcome. For larger API or generator changes, open an issue first so compatibility and generated-code behavior can be discussed before implementation.

Development setup

Install the SDK selected by global.json, then run:

dotnet restore BitsKit.sln --locked-mode
dotnet build BitsKit.sln --configuration Release --no-restore
dotnet test BitsKit.Tests/BitsKit.Tests.csproj --configuration Release --no-build

If a dependency changes, run dotnet restore BitsKit.sln --force-evaluate and commit the updated packages.lock.json files.

For performance work, run a focused benchmark of the affected feature:

./eng/Run-Benchmarks.ps1 -Filter "*MethodName*"

Attach the generated report and environment metadata to the pull request. Use stable local hardware for performance-regression claims.

For optimization work, compare the runtime operations shared with the original repository plus equivalent models compiled by each revision's source generator on the same machine:

./eng/Run-Benchmark-Regression.ps1 -Job Medium -RegressionTolerancePercent 5

The script builds both libraries for net8.0, runs them sequentially under the same .NET 10 BenchmarkDotNet host, and writes Markdown and JSON ratio reports under artifacts/benchmark-regression. Add -Enforce to return a failure when any operation exceeds the allowed slowdown. The Original performance regression workflow exposes the same settings for on-demand hosted runs; use Dry only to validate the harness, not for performance conclusions.

During optimization, use -MethodFilter 'GeneratedAccessor*' (or another PowerShell wildcard) to run a focused comparison before the complete confirmation pass.

Pull requests

  • Add tests for fixes and new behavior.
  • Preserve the public API unless the change is intentionally breaking and documented.
  • Benchmark performance claims using BitsKit.Benchmarks and the original-performance regression harness where applicable.
  • Keep source-generator outputs deterministic and incremental.
  • Update README.md, CHANGELOG.md, and ReleaseNotes.txt when user-facing behavior changes.

By contributing, you agree that your contribution is licensed under the repository's MIT license.