When skiplinks is False (by default), it should as in current implementation, i.e. using stat().

if skiplinks is True, isidr() should use lstat() and same logick. 

If one will be implemented, os.walk() should be patched (see issue12970) to use this new isdir() with this new parameter instead of own logick in os.walk().

If this happens, in keeping with the new os module aesthetic, I suggest the prototype should be

os.isdir(path, *, follow_symlinks=True)

Отправлено с iPad

27.06.2012, в 6:03, Larry Hastings <report@bugs.python.org> написал(а):

No, follow symlinks is not the same. 

Skiplinks mean returning true or false depending on fact that last path element is symlink.
Checking If symlink appear in the middle of the path, this is not issue of this feature request. 

> 
> ----------
> nosy: +larry
> 
> _______________________________________
> Python tracker <report@bugs.python.org>
> <http://bugs.python.org/issue12971>
> _______________________________________

First: there's no chance this will go into 3.1, 3.2, or 3.3 at this point.

Second: I can assure you that switching to the l... version of a function and not following the last symlink is exactly what "follow_symlinks" is for.  Please see the 3.3 documentation on follow_symlinks for more:

http://docs.python.org/3.3/library/os.html#follow-symlinks

Third: I think it's better to keep os.isdir() simple.  Users who want more sophisticated use cases (what to do about symlinks, using a dir_fd parameter) should use os.stat().  So I'm -1 on this.

TBH I'm not sure this even warrants a documentation change ("for more functionality, use os.stat()").