I was looking at Issue 1316 in bench and it looks like it happens because the encoding defaults to latin-1 when not set, causing it to fail when a character outside the 256 range is passed to the subprocess.call() call in the exec_cmd() fn of utils/init.py. Since anyway subprocess is moving towards using run() as the preferred method and that takes a default encoding (similar to the check_output() call in the same file), should we replace all references to subprocess.call() with subprocess.run() with a default utf-8 encoding?