I've run into problems in other repos using the "ubuntu-latest"/"macos-latest" runners, since that changes whenever GitHub bumps the latest. Would it be OK to instead test on the last one or two releases, that is, "ubuntu-22.04"/"ubuntu-20.04" and "macos-12"/"macos-11"? That way there's no surprise CI breakage because of a changing runner.
I've run into problems in other repos using the "ubuntu-latest"/"macos-latest" runners, since that changes whenever GitHub bumps the latest. Would it be OK to instead test on the last one or two releases, that is, "ubuntu-22.04"/"ubuntu-20.04" and "macos-12"/"macos-11"? That way there's no surprise CI breakage because of a changing runner.