But that’s private. I can use std.debug.print directly, but that feels like perhaps the wrong abstraction layer here? Ideally, tests shouldn’t assume that stderr is anything in particular, I would think?
Can you not use std.log for that? I’ve used this in my testing to output more information. You may have to set the std.testing.log_level if you want more detaild logs (default is warn).
Looks like it is expected behavior. It seems like they want users to be able to do the std.log stuff they normally would and have it work with the default test runner. It’s still an open issue, so it’s not fully implemented, but at least confirms the direction.