Developer notes for py_dataset

py_dataset wraps dataset’s libdataset. We try to keep the version of py_dataset in sync with the current release of dataset. libdataset. When a new version of dataset and libdataset are release fetch the libdataset-*.zip archives and unzip them in py_dataset/lib. This should leave you with libdataset.so (Linux), libdataset.dylib (Mac OS X) and libdataset.dll (Windows) as well as libdataset.h (same for all three platforms). You will need to update py_dataset/dataset.py to include any changes you want to reflect in the shared library (e.g. add new functions). Once those twothings have been done you can use python setup.py to install your updates locally and test.

For patches between dataset release we append a period and patch value to the semver that matched dataset’s release. E.g. the first patch to py_dataset paired with dataset 2.2.3 release would be 2.2.3.1. The second 2.2.3.2, etc.

py_dataset is now built/managed using uv.

Release process

• Update shared libraries in py_dataset/lib (e.g. libdataset.so, libdataset.dynlib and libdataset.dll)
• Update py_dataset/dataset.py and py_dataset/libdataset.py as needed
• Update version info in codemeta.json (e.g. cme version releaseNotes dateModified datePublished -e)
• Update README.md and DEVELOPER-NOTES.md
• Test and commit changes to the master branch in the git repo
  • uv run python3 test_dataset.py
• Make a github release if you’re happy with everything
• [optional] Push changes up to pypy