Remove pathlib accessors

[barneygale]

BPO	43012
Nosy	@pitrou, @ericvsmith, @merwok, @encukou, @eryksun, @miss-islington, @barneygale
PRs	bpo-43012: remove pathlib._Accessor #25701

^{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 2022-02-02.15:37:05.067>
created_at = <Date 2021-01-23.18:15:55.455>
labels = ['3.11', 'library', 'performance']
title = 'Remove pathlib accessors'
updated_at = <Date 2022-02-02.15:37:05.063>
user = 'https://github.com/barneygale'

bugs.python.org fields:

activity = <Date 2022-02-02.15:37:05.063>
actor = 'eric.araujo'
assignee = 'none'
closed = True
closed_date = <Date 2022-02-02.15:37:05.067>
closer = 'eric.araujo'
components = ['Library (Lib)']
creation = <Date 2021-01-23.18:15:55.455>
creator = 'barneygale'
dependencies = []
files = []
hgrepos = []
issue_num = 43012
keywords = ['patch']
message_count = 4.0
messages = ['385556', '410732', '410733', '412353']
nosy_count = 7.0
nosy_names = ['pitrou', 'eric.smith', 'eric.araujo', 'petr.viktorin', 'eryksun', 'miss-islington', 'barneygale']
pr_nums = ['25701']
priority = 'normal'
resolution = 'fixed'
stage = 'resolved'
status = 'closed'
superseder = None
type = 'performance'
url = 'https://bugs.python.org/issue43012'
versions = ['Python 3.11']

[barneygale]

Per Pitrou:

The original intent for the “accessor” thing was to have a variant that did all accesses under a filesystem tree in a race condition-free way using openat and friends. It turned out to be much too hairy to actually implement, so was entirely abandoned, but the accessor abstraction was left there.

https://discuss.python.org/t/make-pathlib-extensible/3428/2

Accessors are:

• Lacking any internal purpose - '_NormalAccessor' is the only implementation
• Lacking any firm conceptual difference to Path objects themselves (inc. subclasses)
• Non-public, i.e. underscore prefixed - '_Accessor' and '_NormalAccessor'
• Unofficially used to implement customized Path objects, but once once bpo-24132 lands there will be a supported route for that.

When bpo-24132 lands, I think we're free to remove accessors and simplify some code

[merwok]

PR looks good, just needs a look by a pathlib expert.

[barneygale]

Thank you Éric!

[miss-islington]

New changeset 08f8301 by Barney Gale in branch 'main':
bpo-43012: remove pathlib._Accessor (GH-25701)
08f8301

[ronf]

This change is actually causing a problem for me. I have a unit test case which mocks out os.path.expanduser() but not the pathlib version. This works in 3.10 and earlier due to the indirection of the accessors, but in 3.11 the code ends up failing, since now the pathlib version of expanduser() is also getting into my mock.

I wonder if there might be other code out there which also might be depending on this behavior, without even realizing it.