Our current Dockerfile uses a build arg to create a fake-conditional in the Dockerfile to switch between production/development use cases. This is a hack at best and results in both sides of the conditional being build, regardless of the use case.

We can switch to using the --target flag e.g. docker build --target=production . and, if our Dockerfile has a proper multi-stage dependency graph isolated to the right target, we should not have the problems associated with the build arg hack.