The high level overview is what led me there: https://www.pantsbuild.org/docs/environments#environment-aware-options

I tried that, but I think the problem is, that the environment variable is already defined in

env_vars_devcontainer()

and this overrides whatever value I decalre for it in

test_extra_env_vars

.

If so, that sounds like a bug. If you drop use of environments (just using local is like not using them), it should be the case that plain old test extra_env_vars overrides subprocess_environment env_vars.

Here ist says that: "Any fields that are defined in that environment will override the values from options set by

pants.toml

, command line values, or environment variables." To me this sounds like the environment variables defined in my "local" environment will override anything that is set somewhere else. Or am I misinterpreting this sentence? But if this is indeed a bug, I will file an issue.

I don't think I'm being clear. So forget environments for a second. In plain old pants you have (for Python)

[subprocess-environment] env_vars

which applies to all Python subprocess and

[test] extra_env_vars

which overlays that but just for test subprocesses. Now pulling environments back in, the local environment has a mirror of both of those same options. Using the equivalent pair of options there should have the equivalent overlay effect.

I have more than one, since one test needs to run in its own docker container, therefore, I think the usage of environments makes sense for me. I just wanted to give a simple overview what I wanted to to so I left out this detail.

Ok, great.