Thanks to the hard work in #58, we can spin up dockerized Spark for integration tests. I think all PRs (including ones from external contributors) should run unit and integration tests in Circle.
Releases
For now, I think a good release flow might look like:
dbt-spark maintainer cuts a new GitHub release off the master branch, tagged with a semantic version
Automated deployment of that versioned release to pypi (+ homebrew?) via GitHub Actions
Eventually we may find we need dev/version branches, alpha/beta/release candidates, etc.
Addressed to the best of our current ability in #62 + #72. As our release cadence for dbt-core improves (bumpversion --> GitHub tag --> pypi), the plugins can follow.
CI
Thanks to the hard work in #58, we can spin up dockerized Spark for integration tests. I think all PRs (including ones from external contributors) should run unit and integration tests in Circle.
Releases
For now, I think a good release flow might look like:
dbt-spark
maintainer cuts a new GitHub release off themaster
branch, tagged with a semantic versionEventually we may find we need
dev/version
branches, alpha/beta/release candidates, etc.