New submission from Andrei Kulakov <andrei....@gmail.com>:
(note the actual relevant code is in posixmodule.c) os.stat() error can be confusing and misleading when given an fd and with follow_symlinks=False: ValueError: stat: cannot use fd and follow_symlinks together It's less bad when os.stat() is used directly because the user would look at the signature and would have provided the follow_symlinks=False directly, but it's confusing when used indirectly by other function. I've ran into this when reviewing https://github.com/python/cpython/pull/27524 list(os.fwalk(1)) /usr/local/Cellar/python@3.9/3.9.1/Frameworks/Python.framework/Versions/3.9/lib/python3.9/os.py in fwalk(top, topdown, onerror, follow_symlinks, dir_fd) 467 # lstat()/open()/fstat() trick. 468 if not follow_symlinks: --> 469 orig_st = stat(top, follow_symlinks=False, dir_fd=dir_fd) 470 topfd = open(top, O_RDONLY, dir_fd=dir_fd) 471 try: ValueError: stat: cannot use fd and follow_symlinks together ---- A few things are confusing here: I did not use follow_symlinks argument; I can see from traceback that the arg is used but set to False, which is the usual meaning of "do not use follow_symlinks". In addition, fwalk() can probably check that `top` arg is a string and raise an error stating that it should be a string if it's not. If that's done, this issue will no longer happen for current code anywhere in os module, but stat(follow_symlinks=False) is also used in shutil and pathlib (I didn't check if fd may be passed in those cases), but also in 3rd party libraries. So I think it would be clearer to rephrase the error to say stat: cannot use fd and follow_symlinks set to False or NULL together (adding NULL as it may be used from C code). ---------- components: C API, Library (Lib) messages: 399237 nosy: andrei.avk priority: normal severity: normal status: open title: misleading error about fd / follow_symlinks from os.stat() type: behavior versions: Python 3.11 _______________________________________ Python tracker <rep...@bugs.python.org> <https://bugs.python.org/issue44868> _______________________________________ _______________________________________________ Python-bugs-list mailing list Unsubscribe: https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com
