[issue36656] Race conditions due to os.link and os.symlink POSIX specification

Tom Hale report at bugs.python.org
Mon Jun 3 06:36:04 EDT 2019


Tom Hale <tomnott at gmail.com> added the comment:

Serhiy wrote

> Detected problem is better than non-detected problem.

I agree. I also assert that no problem (via a shutil wrapper) is better than a detected problem which may not be handled.

While it's up to the programmer to handle exceptions, it's only systems programmers who will realise that there is a sometimes-occurring exception which should be handled.

So most people won't handle it. They don't know that they need to.

The shutil os.link and os.symlink wrappers under discussion on python-ideas prevent non-system-programmers from needing to know about the race-condition case (and correctly handling it).

If something is difficult to get right, then let's put it in shutil and save everyone reinventing the wheel.

3 specific cases in which an atomic link replacement would be useful are listed here:

https://code.activestate.com/lists/python-ideas/56195/

----------

_______________________________________
Python tracker <report at bugs.python.org>
<https://bugs.python.org/issue36656>
_______________________________________


More information about the Python-bugs-list mailing list