From 5a0e31436b2f3f84239d4b3993bc631660319fb4 Mon Sep 17 00:00:00 2001
From: Athanasius <github@miggy.org>
Date: Wed, 7 Sep 2022 12:27:35 +0100
Subject: [PATCH] tests: journal_lock: Works 100% on Linux, remove 'plan'
docstring text
---
tests/journal_lock.py/test_journal_lock.py | 39 ++++------------------
1 file changed, 6 insertions(+), 33 deletions(-)
diff --git a/tests/journal_lock.py/test_journal_lock.py b/tests/journal_lock.py/test_journal_lock.py
index 7afa588c..016a8f1b 100644
--- a/tests/journal_lock.py/test_journal_lock.py
+++ b/tests/journal_lock.py/test_journal_lock.py
@@ -1,33 +1,4 @@
-"""
-Tests for journal_lock.py code.
-
-Tests:
- - Is file actually locked after obtain_lock(). Problem: We opened the
- file in a manner which means nothing else can open it. Also I assume
- that the same process will either be allowed to lock it 'again' or
- overwrite the lock.
-
- Expected failures if:
-
- 1. Lock already held (elsewhere).
- 2. Can't open lock file 'w+'.
- 3. Path to lock file doesn't exist.
- 4. journaldir is None (default on Linux).
-
- - Does release_lock() work? Easier to test, if it's worked....
- 1. return True if not locked.
- 2. return True if locked, but successful unlock.
- 3. return False otherwise.
-
- - JournalLock.set_path_from_journaldir
- 1. When journaldir is None.
- 2. Succeeds otherwise?
-
- - Can any string to pathlib.Path result in an invalid path for other
- operations?
-
- - Not sure about testing JournalAlreadyLocked class.
-"""
+"""Tests for journal_lock.py code."""
import multiprocessing as mp
import os
import pathlib
@@ -327,11 +298,13 @@ class TestJournalLock:
# Now attempt to lock with to-test code
jlock = JournalLock()
second_attempt = jlock.obtain_lock()
- # Fails on Linux, because flock(2) is per process, so we'd need to
- # use multiprocessing to test this.
assert second_attempt == JournalLockResult.ALREADY_LOCKED
- # And need to release any handles on the lockfile
+
+ # Need to release any handles on the lockfile else the sub-process
+ # might not be able to clean up properly, and that will impact
+ # on later tests.
jlock.journal_dir_lockfile.close()
+
print('Telling sub-process to quit...')
exit_q.put('quit')
print('Waiting for sub-process...')