bpo-41357: Add a sentence to os.path.abspath() clarifying that it pre… #21596
Conversation
…serves symlink names. Also, replace "os.path.abspath" with "os.path.realpath" in the table at the end of the Path class section.
Doc/library/os.path.rst
Outdated
| @@ -65,7 +65,8 @@ the :mod:`glob` module.) | |||
|
|
|||
| Return a normalized absolutized version of the pathname *path*. On most | |||
| platforms, this is equivalent to calling the function :func:`normpath` as | |||
| follows: ``normpath(join(os.getcwd(), path))``. | |||
| follows: ``normpath(join(os.getcwd(), path))``. Unlike :func:`realpath`, | |||
| this function preserves symlink names. | |||
There was a problem hiding this comment.
I think this should be phrased in the negative that "this function does not resolve symlink targets".
If the reader follows through to the documentation of normpath, it's clear that it resolves ".." components, which may remove symlinks from the path. But based on just this sentence, it's not clear what it means to preserve symlinks. A different implementation might retain all ".." components, as pathlib's absolute method does. But the current implementation can change the real target of a Unix path, e.g. normpath('spam/symlink/../eggs') -> "spam/eggs".
In Windows, this is usually not a problem since the Windows API works like this anyway. The one exception is in the target of a relative symlink, e.g. symlink(r'spam\symlink\..\eggs', 'eggs') is not the same as symlink(r'spam\eggs', 'eggs').
There was a problem hiding this comment.
I updated the wording to more closely reflect, in a negative fashion, the language used for realpath().
…serves symlink names. Also, replace "os.path.abspath" with "os.path.realpath" in the table at the end of the Path class section.
https://bugs.python.org/issue41357