Our production deployments run on postgres, but all our tests run on sqlite. These two are different enough to make our tests (and thus CI results) not necessarily reflective of the behavior of our production deployment.
What's the solution?
Using the pgtemp crate to create a postgres instance for unit tests to use. It uses a local installation of postgres instead of installing docker and then downloading postgres, which would make CI take much longer.
This PR also removes all mentions of sqlite from the repo, and adds instructions for quickly spinning up a local postgres instance via docker.
What's the problem?
Our production deployments run on postgres, but all our tests run on sqlite. These two are different enough to make our tests (and thus CI results) not necessarily reflective of the behavior of our production deployment.
What's the solution?
Using the
pgtempcrate to create a postgres instance for unit tests to use. It uses a local installation of postgres instead of installing docker and then downloading postgres, which would make CI take much longer.This PR also removes all mentions of sqlite from the repo, and adds instructions for quickly spinning up a local postgres instance via docker.