pathlib: remove partial support for preserving accessor when modifying a path

[barneygale]

BPO	40038
Nosy	@zooba, @barneygale
PRs	bpo-40038: pathlib: remove partial support for preserving accessor when modifying a path #19342

^{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 2021-04-07.11:35:15.242>
created_at = <Date 2020-03-22.05:15:29.994>
labels = ['library', '3.9', 'performance']
title = 'pathlib: remove partial support for preserving accessor when modifying a path'
updated_at = <Date 2021-04-07.11:35:15.241>
user = 'https://github.com/barneygale'

bugs.python.org fields:

activity = <Date 2021-04-07.11:35:15.241>
actor = 'steve.dower'
assignee = 'none'
closed = True
closed_date = <Date 2021-04-07.11:35:15.242>
closer = 'steve.dower'
components = ['Library (Lib)']
creation = <Date 2020-03-22.05:15:29.994>
creator = 'barneygale'
dependencies = []
files = []
hgrepos = []
issue_num = 40038
keywords = ['patch']
message_count = 2.0
messages = ['364783', '390392']
nosy_count = 2.0
nosy_names = ['steve.dower', 'barneygale']
pr_nums = ['19342']
priority = 'normal'
resolution = 'fixed'
stage = 'resolved'
status = 'closed'
superseder = None
type = 'performance'
url = 'https://bugs.python.org/issue40038'
versions = ['Python 3.9']

[barneygale]

pathlib.Path._init() accepts a 'template' argument that pathlib uses - in some cases - to preserve the current accessor object when creating modified path objects. This works for resolve(), absolute() and readlink(), but no other cases!

As customizing the accessor is not something we support (yet! see https://discuss.python.org/t/make-pathlib-extensible/3428), and the majority of path methods do not call _init() with a 'template' argument (and so do not preserve the accessor), I suggest this internal functionality be removed. Together with bpo-39682 / #63046, I believe this would allow us to remove Path._init() entirely, which would be a small performance win and a simplification of the code.

Demo:

import pathlib


class CustomAccessor(pathlib._NormalAccessor):
    pass


def print_accessor(path):
    if isinstance(path._accessor, CustomAccessor):
        print("    %r: custom" % path)
    else:
        print("    %r: normal" % path)


print("Here's a path with a custom accessor:")
p = pathlib.Path("/tmp")
p._accessor = CustomAccessor()
print_accessor(p)

print("Our accessor type is retained in resolve(), absolute() and readlink():")
print_accessor(p.absolute())
print_accessor(p.resolve())
#print_accessor(p.readlink())

print("But not in any other path-creating methods!")
print_accessor(p.with_name("foo"))
print_accessor(p.with_suffix(".foo"))
print_accessor(p.relative_to("/"))
print_accessor(p / "foo")
print_accessor(p.joinpath("foo"))
print_accessor(p.parent)
print_accessor(p.parents[0])
print_accessor(list(p.iterdir())[0])
print_accessor(list(p.glob("*"))[0])
print_accessor(list(p.rglob("*"))[0])
#print_accessor(p.expanduser())

[zooba]

New changeset 2219187 by Barney Gale in branch 'master':
bpo-40038: pathlib: remove partial support for preserving accessor when modifying a path (GH-19342)
2219187