isinstance checks in os.path.normcase redundant with os.fspath

[wm75]

BPO	30427
Nosy	@brettcannon, @serhiy-storchaka, @wm75, @miss-islington, @isidentical
PRs	bpo-30427: eliminate redundant type checks in os.path.normcase() #1712

^{Note: these values reflect the state of the issue at the time it was migrated and might not reflect the current state.}

Show more details

GitHub fields:

assignee = None
closed_at = <Date 2020-01-02.22:13:44.727>
created_at = <Date 2017-05-22.08:17:57.051>
labels = ['3.7', 'library', 'performance']
title = 'isinstance checks in os.path.normcase redundant with os.fspath'
updated_at = <Date 2020-01-02.22:13:44.726>
user = 'https://github.com/wm75'

bugs.python.org fields:

activity = <Date 2020-01-02.22:13:44.726>
actor = 'brett.cannon'
assignee = 'none'
closed = True
closed_date = <Date 2020-01-02.22:13:44.727>
closer = 'brett.cannon'
components = ['Library (Lib)']
creation = <Date 2017-05-22.08:17:57.051>
creator = 'wolma'
dependencies = []
files = []
hgrepos = []
issue_num = 30427
keywords = []
message_count = 7.0
messages = ['294130', '294131', '294137', '294206', '339077', '358868', '359223']
nosy_count = 5.0
nosy_names = ['brett.cannon', 'serhiy.storchaka', 'wolma', 'miss-islington', 'BTaskaya']
pr_nums = ['1712']
priority = 'normal'
resolution = 'fixed'
stage = 'resolved'
status = 'closed'
superseder = None
type = 'performance'
url = 'https://bugs.python.org/issue30427'
versions = ['Python 3.6', 'Python 3.7']

[wm75]

os.path.normcase as defined in both posixpath and ntpath is now calling os.fspath on its argument first. With that I think the following isinstance(str, bytes) checks have become redundant since AFAIU os.fspath is guaranteed to return either str or bytes instances.

[wm75]

Just created a PR for this, which eliminates the redundancy. This also changes the error message (making it less specific), but not the type of a raised exception.
If you think that the error message deserves to be preserved that could, of course, be done too.

[serhiy-storchaka]

Does calling os.fspath() needed in normcase() at all? Is normcase() even called with arguments which are not just str/bytes?

[brettcannon]

The problem with leaving os.fspath() out of os.path.normcase() is that suddenly a single function that deals with paths won't work with path-like objects. So that means support for path-like objects won't implicitly work in code that assumes a path but doesn't explicitly go out of its way to support path-like objects.

[miss-islington]

New changeset 74510e2 by Miss Islington (bot) (Wolfgang Maier) in branch 'master':
bpo-30427: eliminate redundant type checks in os.path.normcase() (GH-1712)
74510e2

[isidentical]

Looks like there is nothing left after PR 1712 if I am not missing something, can this issue be closed?

[brettcannon]

Yep, it looks like this is fixed, Batuhan. Thanks for letting us know!