Steve Dower <steve.do...@python.org> added the comment:
> I've pasted the diff below because I'm not yet convinced that it is correct > (in particular the value for "argv0".) argv0 is literally what C sees in argv[0], which in the framework case I believe is calculated by a launcher? The getpath.py change is probably adding an OS check for this line: https://github.com/python/cpython/blob/main/Modules/getpath.py#L304 if ENV_PYTHONEXECUTABLE or ENV___PYVENV_LAUNCHER__: # If set, these variables imply that we should be using them as # sys.executable and when searching for venvs. However, we should # use the argv0 path for prefix calculation if os_name != 'darwin': base_executable = executable if not real_executable: real_executable = executable executable = ENV_PYTHONEXECUTABLE or ENV___PYVENV_LAUNCHER__ executable_dir = dirname(executable) I think the comment "we should use the argv0 path for prefix calculation" isn't correct (anymore?) either, since updating executable_dir is going to do the opposite of that. But I guess there were tests proving otherwise. It won't affect this case, since both argv0 and ENV___PYVENV_LAUNCHER__ appear to be under the same prefix (at least in Ronald's example). ---------- _______________________________________ Python tracker <rep...@bugs.python.org> <https://bugs.python.org/issue46890> _______________________________________ _______________________________________________ Python-bugs-list mailing list Unsubscribe: https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com
