Releasing New Version
DataJunction project publishes all of its backend services and the Python client library to PyPI:
- datajunction-clients/python
- datajunction-query
- datajunction-server
- datajunction-reflection
Javascript client and the UI component go to NPM:
- datajunction-clients/javascript
- datajunction-ui
JAVA client (TODO):
- datajunction-clients/java
Documentation (TODO).
- docs
Dev Releases
If you’d like make a dev release of any DataJunction component switch to a corresponding component directory and run:
make dev-release
Look for the version and a target release url in the output. It will differ between the languages we use.
Example from Pyhton client:
% make dev-release
hatch version dev
Old: 0.0.1a20
New: 0.0.1a20.dev0
hatch build
[sdist]
dist/datajunction-0.0.1a20.dev0.tar.gz
[wheel]
dist/datajunction-0.0.1a20.dev0-py3-none-any.whl
hatch publish
dist/datajunction-0.0.1a20.dev0.tar.gz ... success
dist/datajunction-0.0.1a20.dev0-py3-none-any.whl ... success
[datajunction]
https://pypi.org/project/datajunction/0.0.1a20.dev0/
Official Releases
In order to simplify version management we decided to release all the components together. Releasing all components with the same version carries some challanges with it, but we we’ll try to use this approach for some time and reconsider if necessary.
Three steps:
To make a new official release for all the components simply run the Version Bump Github action with the default settings. The default setting will make sure you are staying with the current version cycle, which at the moment is ALPHA.
Once the above workflow runs succesfully we should recive a PR for the version updates. This will allow us to double-check the versions before the final release.
After the above PR is merged a Version Publish’ing Github action will build and publish all the versions to its corresponding repositories.