tools/README.md
To download CI build artifacts from S3, set up your AWS credentials
and run tools/download-s3.py <tag|sha1>, the artifacts will be downloaded
into dl-<tag|sha1>.
To upload binary packages to PyPi (test.pypi.org in this example to be safe), do:
$ twine upload -r test dl-<tag|sha1>/*
Run tools/generate_kafka_error_stub_codes.py to generate the list of error codes
supported by the compiled C extension.
Running it with --check compares the error code attributes in KafkaError stub class
in src/confluent_kafka/cimpl.pyi with the error codes supported by librdkafka
This script should be used to update the KafkaError stub class with the up-to-date
error codes supported by librdkafka
To build the manylinux Python packages follow these steps:
NOTE: Docker is required.
Install cibuildwheel:
$ pip install -r tools/requirements-manylinux.txt
Build using cibuildwheel:
$ tools/cibuildwheel-build.sh wheelhouse
To skip Python platform configurations, use glob matching in CIBW_SKIP env. This example only builds for Python 2.7 x64:
$ CIBW_SKIP="cp3* cp*i686*" tools/cibuildwheel-build.sh wheelhouse
Packages will now be available in wheelhouse/
After wheels have been downloaded with download-s3.py, perform smoketests
by running tools/test-wheels.sh <download-dir>, e.g.,
tools/test-wheels.sh tools/dl-v1.5.0rc1.
This script preferably be run on OSX (with Docker installed) so that
both OSX and Linux wheels are tested.