Test test_script_helper fails, details:

Running Debug|Win32 interpreter...
== CPython 3.8.0a0 (heads/master:6cdb7954b0, Feb 24 2018, 17:25:46) [MSC v.1912 32 bit (Intel)]
== Windows-10-10.0.16299-SP0 little-endian
== cwd: \cpython\build\test_python_7920
== CPU count: 4
== encodings: locale=cp1252, FS=utf-8
Run tests sequentially
0:00:00 [1/1] test_script_helper
test_assert_python_failure (test.test_script_helper.TestScriptHelper) ... ok
test_assert_python_failure_raises (test.test_script_helper.TestScriptHelper) ... ok
test_assert_python_isolated_when_env_not_required (test.test_script_helper.TestScriptHelper) ... ok
test_assert_python_not_isolated_when_env_is_required (test.test_script_helper.TestScriptHelper)
Ensure that -I is not passed when the environment is required. ... ok
test_assert_python_ok (test.test_script_helper.TestScriptHelper) ... ok
test_assert_python_ok_raises (test.test_script_helper.TestScriptHelper) ... ok
test_interpreter_requires_environment_details (test.test_script_helper.TestScriptHelperEnvironment) ... FAIL
test_interpreter_requires_environment_false (test.test_script_helper.TestScriptHelperEnvironment) ... FAIL
test_interpreter_requires_environment_true (test.test_script_helper.TestScriptHelperEnvironment) ... FAIL

======================================================================
FAIL: test_interpreter_requires_environment_details (test.test_script_helper.TestScriptHelperEnvironment)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "C:\Users\andor\Documents\Projects\cpython\lib\unittest\mock.py", line 1191, in patched
    return func(*args, **keywargs)
  File "C:\Users\andor\Documents\Projects\cpython\lib\test\test_script_helper.py", line 101, in test_interpreter_requires_environment_details
    self.assertFalse(script_helper.interpreter_requires_environment())
AssertionError: True is not false

======================================================================
FAIL: test_interpreter_requires_environment_false (test.test_script_helper.TestScriptHelperEnvironment)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "C:\Users\andor\Documents\Projects\cpython\lib\unittest\mock.py", line 1191, in patched
    return func(*args, **keywargs)
  File "C:\Users\andor\Documents\Projects\cpython\lib\test\test_script_helper.py", line 95, in test_interpreter_requires_environment_false
    self.assertFalse(script_helper.interpreter_requires_environment())
AssertionError: True is not false

======================================================================
FAIL: test_interpreter_requires_environment_true (test.test_script_helper.TestScriptHelperEnvironment)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "C:\Users\andor\Documents\Projects\cpython\lib\unittest\mock.py", line 1191, in patched
    return func(*args, **keywargs)
  File "C:\Users\andor\Documents\Projects\cpython\lib\test\test_script_helper.py", line 89, in test_interpreter_requires_environment_true
    self.assertEqual(1, mock_check_call.call_count)
AssertionError: 1 != 0

----------------------------------------------------------------------
Ran 9 tests in 0.240s

FAILED (failures=3)
test test_script_helper failed
test_script_helper failed

1 test failed:
    test_script_helper

Total duration: 281 ms
Tests result: FAILURE

For reference this is happening with a clean checkout and build of master for me, which seems like it shouldn't be the case...

I confirm that I reproduce the issue on Windows when running "python -m test test_script_helper -v" from the source code directory, on 3.6, 3.7 and master branches.

Using Python 3.6 installed on the system (using python.org installer), the same command pass.

Ditto with fresh update and build, except that I used installed 3.7.0rc1 (and it passed).  This is not regularly failing on the (Windows) buildbots, so there is something 'different' about the repository on some non-buildbot user machines.

I am pretty sure that I have run the full suite OK within the last month, certainly since last February.  The test has not been touched in 3 years.

I confirm the issue too.
I am running the tests on a win 10 Ver 1709, clean build.

Most likely the issue is caused by the changes from commit fa9a502a387a26ba116840d35ae4a30f672e78ad
at test.support.script_helper
The function in test was altered to support a different behaviour when run from CI. The related tests were not changed.

As I am relativly new to python-dev: Does the bot/ user machines differ in the existence of PYTHONHOME? (Then the passing of the bot runs are plausible)

The tests are modified/ one added. PR attached.

f:\dev\3x>python -m test test_script_helper
...
test_script_helper failed

f:\dev\3x> git pr 8034
...
Switched to branch 'pr_8034'

f:\dev\3x>python -m test test_script_helper
...
== Tests result: SUCCESS ==

Lorenz, welcome!  Please sign CLA so we can review and merge this.

New changeset a390cb6b6fb6638e91d699513167e8f35b47f846 by Victor Stinner (Lorenz Mende) in branch 'master':
bpo-32942: Fix environment dependent test_script_helper (GH-8034)
https://github.com/python/cpython/commit/a390cb6b6fb6638e91d699513167e8f35b47f846

New changeset 6b3ff35670869fb915d43ac434c473c1fc11598e by Miss Islington (bot) in branch '3.7':
bpo-32942: Fix environment dependent test_script_helper (GH-8034)
https://github.com/python/cpython/commit/6b3ff35670869fb915d43ac434c473c1fc11598e

New changeset aea19a0530726c7ba6a9461b0024506529dc35e8 by Miss Islington (bot) in branch '3.6':
bpo-32942: Fix environment dependent test_script_helper (GH-8034)
https://github.com/python/cpython/commit/aea19a0530726c7ba6a9461b0024506529dc35e8

Thank you Lorenz Mende for your fix, and Terry for your review! I tested manually the fix, and I confirm that it works (Terry also tested and confirmed that the fix works as expected).

I applied your fix to master, and backport it to 3.6 and 3.7. Python 2.7 has no test on script_helper, so it's not affected by this bug.