Does not work with vscode unittest runner?

[ziesemer]

Describe the bug

When running tests using with the VS Code unittest runner, test failures no longer report properly. Seems to be an issue with the pyfakefs setup intercepting calls from the test runner itself, possibly even outside the scope of a running test?

How To Reproduce

Use the same code as from https://pytest-pyfakefs.readthedocs.io/en/latest/usage.html#patch-using-fake-filesystem-unittest, but with an added assertion at the end to ensure the test fails:

import os
import typing

from pyfakefs.fake_filesystem_unittest import TestCase

class ExampleTestCase(TestCase):
	@typing.override
	def setUp(self):
		self.setUpPyfakefs()

	def test_create_file(self):
		file_path = "/test/file.txt"
		self.assertFalse(os.path.exists(file_path))
		self.fs.create_file(file_path)
		self.assertTrue(os.path.exists(file_path))
		assert(False)

The test result outputs show:

test_create_file (tests.test_pyfakefs.ExampleTestCase.test_create_file) ... FAIL
Error attempting to connect to extension named pipe /run/user/1000/python-test-results-567064382b9f3143b111[vscode-unittest]: [Errno 2] No such file or directory: '/run/user/1000'

And the test UI only shows this test as a skipped test, not a failure.

I've played with this quite a bit, trying to only include or exclude certain paths, changing where the setup happens and is torn down, etc., and I'm not seeming to find a workable solution.

It seems that pyfakefs and https://github.com/microsoft/vscode-python/blob/main/python_files/unittestadapter/pvsc_utils.py are probably not getting along somewhere?

Your environment

Linux-6.12.62+rpt-rpi-2712-aarch64-with-glibc2.41
Python 3.14.3 (main, Feb 8 2026, 03:02:41) [GCC 14.2.0]
pyfakefs 6.1.1
Visual Studio Code 1.1.09.5

[mrbean-bremen]

Thanks! I'm usually not working with VSCode, so I didn't notice this. I know that it did work with an earlier version, so this is either due to a change in the debugger, or in pyfakefs itself.
Problems with test runners or debuggers have happened before, because debuggers often rely on the real filesystem during test, when the filesystem is mocked.
I will see if I can reproduce this and will try to find a workaround.

[mrbean-bremen]

This is weird. I first thought that I can't reproduce the problem, because I adapted existing tests to reproduce the problem.
Then I used exactly your test and could reproduce it, both under Windows and Linux.
Turns out, this only happens if the failing test is the first that is run in the test class (e.g. the first alphabetically). If the first test passes, the rest of the tests behaves as expected, e.g. this works:

class ExampleTestCase(TestCase):
    @typing.override
    def setUp(self):	
        self.setUpPyfakefs()

    def test_a(self):
        pass

    def test_create_file(self):
        ...
        assert(False)

This even works if the first test is skipped:

class ExampleTestCase(TestCase):
    def setUp(self):	
        self.setUpPyfakefs()

    @unittest.skip("skipping")
    def test_a(self):
        pass

    def test_b(self):
        assert(False)

I'm not sure yet what causes this yet - this may take some time...

[ziesemer]

That is crazy! Thanks for the help!