Prevent site-packages .pth files from causing test_site failure: test_site.test_startup_imports() failure

[Rosuav]

BPO	27807
Nosy	@terryjreedy, @vstinner, @ezio-melotti, @voidspace, @Rosuav, @miss-islington
PRs	bpo-27807: Skip test_site.test_startup_imports() if pth file #19060 [3.8] bpo-27807: Skip test_site.test_startup_imports() if pth file (GH-19060) #19089 [3.8] bpo-27807: Skip test_site.test_startup_imports() if pth file (GH-19060) #19090 [3.7] [3.8] bpo-27807: Skip test_site.test_startup_imports() if pth file (GH-19060) (GH-19090) #19093
Files	no-site-packages-in-test.patch

^{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 2020-03-20.14:32:53.418>
created_at = <Date 2016-08-19.22:05:23.814>
labels = ['3.7', '3.8', 'type-bug', 'tests', '3.9']
title = 'Prevent site-packages .pth files from causing test_site failure: test_site.test_startup_imports() failure'
updated_at = <Date 2020-03-20.14:32:53.416>
user = 'https://github.com/Rosuav'

bugs.python.org fields:

activity = <Date 2020-03-20.14:32:53.416>
actor = 'vstinner'
assignee = 'none'
closed = True
closed_date = <Date 2020-03-20.14:32:53.418>
closer = 'vstinner'
components = ['Tests']
creation = <Date 2016-08-19.22:05:23.814>
creator = 'Rosuav'
dependencies = []
files = ['44158']
hgrepos = []
issue_num = 27807
keywords = ['patch']
message_count = 8.0
messages = ['273160', '273165', '364523', '364526', '364536', '364668', '364672', '364673']
nosy_count = 6.0
nosy_names = ['terry.reedy', 'vstinner', 'ezio.melotti', 'michael.foord', 'Rosuav', 'miss-islington']
pr_nums = ['19060', '19089', '19090', '19093']
priority = 'normal'
resolution = 'fixed'
stage = 'resolved'
status = 'closed'
superseder = None
type = 'behavior'
url = 'https://bugs.python.org/issue27807'
versions = ['Python 3.7', 'Python 3.8', 'Python 3.9']

[Rosuav]

Having matplotlib installed globally causes test_site to fail, due to the .pth file importing types. The point of the test is to monitor startup dependencies (and thus time) for the core interpreter and standard library, not third-party libraries, so having a test failure due to a .pth is not overly useful.

Solution suggested by Steve Dower: hack out all .pth handling during this one test, leaving everything else unchanged.

[terryjreedy]

I think hitting adventurous users like Chris who locally compile and test with this failure is wrong. There is no core dev 'site' expert, so I picked 'testing'.

[vstinner]

Fedora and RHEL downstream issue: https://bugzilla.redhat.com/show_bug.cgi?id=1814392

[vstinner]

I marked bpo-35691 as a duplicate of this issue.

[vstinner]

New changeset d18de46 by Victor Stinner in branch 'master':
bpo-27807: Skip test_site.test_startup_imports() if pth file (GH-19060)
d18de46

[vstinner]

New changeset ba26bf3 by Victor Stinner in branch '3.8':
[3.8] bpo-27807: Skip test_site.test_startup_imports() if pth file (GH-19060) (GH-19090)
ba26bf3

[miss-islington]

New changeset 6056b7b by Miss Islington (bot) in branch '3.7':
[3.8] bpo-27807: Skip test_site.test_startup_imports() if pth file (GH-19060) (GH-19090)
6056b7b

[vstinner]

The issue is now fixed in 3.7, 3.8 and master branches.

Thanks Chris Angelico for the bug report.