FastPath.zip_children can give duplicate results on Python 3.8

[jaraco]

BPO	41855
Nosy	@warsaw, @jaraco
PRs	(merge after 3.8.6 final) [3.8] bpo-41855: Fix duplicate results in FastPath.zip_children() #22403 [3.8] bpo-41855: Fix duplicate results in FastPath.zip_children() #22404

^{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 = 'https://github.com/jaraco'
closed_at = <Date 2021-05-27.01:12:30.519>
created_at = <Date 2020-09-24.20:30:47.380>
labels = ['3.8', 'type-bug', 'library']
title = 'FastPath.zip_children can give duplicate results on Python 3.8'
updated_at = <Date 2021-05-27.01:12:30.519>
user = 'https://github.com/jaraco'

bugs.python.org fields:

activity = <Date 2021-05-27.01:12:30.519>
actor = 'jaraco'
assignee = 'jaraco'
closed = True
closed_date = <Date 2021-05-27.01:12:30.519>
closer = 'jaraco'
components = ['Library (Lib)']
creation = <Date 2020-09-24.20:30:47.380>
creator = 'jaraco'
dependencies = []
files = []
hgrepos = []
issue_num = 41855
keywords = ['patch']
message_count = 3.0
messages = ['377461', '377462', '378699']
nosy_count = 2.0
nosy_names = ['barry', 'jaraco']
pr_nums = ['22403', '22404']
priority = 'normal'
resolution = 'fixed'
stage = 'resolved'
status = 'closed'
superseder = None
type = 'behavior'
url = 'https://bugs.python.org/issue41855'
versions = ['Python 3.8']

[jaraco]

Scott J. reports in an e-mail:

When upgrading to Python 3.8.2, we noticed two issues in
importlib.metadata which are not present in importlib_metadata.

1. FastPath.zip_children() is very slow for large zips. Python 3.8.3
   > already has a fix for this
   > ([bug]{.ul},
   > [changelog]{.ul}),
   > which includes merging in importlib_metadata 1.5.0.

2. FastPath.zip_children() can give duplicate results, causing
   > duplicate results in entry_points(). This issue was reported in
   > importlib_metadata in March
   > ([link]{.ul})
   > and fixed in importlib_metadata 1.5.2
   > ([changelog]{.ul}).

In June, importlib_metadata 1.6.1 (including the fix for \#2) was merged
into Python
([bug]{.ul},
[PR]{.ul})
and backported to 3.9
([PR]{.ul}).
However, the backport for Python 3.8 was **not** merged
([PR]{.ul}).
Jason said:

Of course this can\'t backport to 3.8; importlib.resources.files
doesn\'t exist there. No problem. Bug fixes will need to be backported
specially if needed.

As far as I can tell, no version of importlib_metadata newer than 1.5.0
has ever been merged into Python 3.8, and so bug \#2 is still present
(tested in 3.8.6rc1). This seems like it falls under the \"bug fixes if
needed\" category. We just need to fix the duplicate entries, as
reported in importlib_metadata. This should be trivial. The diff in
python 3.9 is
[here]{.ul}.

[jaraco]

The relevant commit is already present in importlib_metadata as 079ca1cb701a5f4aab0a9cb00b0782c7ea8fb70b.

[jaraco]

New changeset 967fdda by Jason R. Coombs in branch '3.8':
[3.8] bpo-41855: Fix duplicate results in FastPath.zip_children() (bpo-22404)
967fdda