PyPI Rebuild Study Dataset and Sampled Artifacts

This dataset accompanies the paper "No Snake Oil: Verifying Python Package Builds". It contains the package/version input dataset, experiment-level rebuild reports, and a reduced sample of raw rebuild artifacts from the PyPI rebuild study. The full rebuild artifact tree is substantially larger than Zenodo's standard upload limit. Therefore, this deposit includes the complete small metadata/dataset archive together with a reduced artifact subset. The uploaded artifact subset consists of the first five 500 MB artifact archives and covers 87 package-version artifact directories, with approximately 2.33 GB of artifact content before gzip/archive overhead. `zenodo-metadata.tar.gz` contains the dataset files, retained logs, manifests, and metadata. `ARTIFACT_PARTS.csv` maps each included package/version artifact directory to the archive that contains it. The artifact archives `zenodo-artifacts-500mb-part-001.tar.gz` through `zenodo-artifacts-500mb-part-005.tar.gz` are independently extractable; extracting them into the same directory as the metadata archive reconstructs the deposited subset of the `sample_artifacts/` tree. The experimental and post-analysis scripts are archived separately in the project GitHub repository.