The test test_build_reply_arguments_chat_id_and_message_id was failing randomly, mostly when using pytest --lf (about 1 in 10 times). The reason was message.chat.username being set to 'username', so setting it to None will solve this problem.

Updated 3 tests of test_message.py.

b64a9b0

The tests test_link_with_topics and test_link_with_reply were failing if beeing runned alone ( since message.chat.type was private).The test test_build_reply_arguments_chat_id_and_message_id was failing randomly, mostly when using pytest --lf (about 1 in 10 times). The reason was message.chat.username being set to 'username', so setting it to None will solve this problem.

harshil21 requested changes

Aug 16, 2024

View reviewed changes

Copy link

Member

harshil21 left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others.Learn more.

Thanks for the fix! I noticed this myself a couple of times before but didn't investigate the cause.

To elaborate further after investigation - part of the reason why it doesn't fail when the whole test file is run is because themessage fixture is module scoped, and other tests are modifying that value (bad practice imo) which leads to a slightly different state for the next test. They fail individually because we initialize themessage fixture with a private chat anyway, which doesn't have public links obviously.

I would suggest changing the scope of that fixture back to "function" (the default) to reduce the chance of this occuring in the future.

harshil21 added the ⚙️ testsaffected functionality: tests label

Aug 16, 2024

Changed pytest scope of the function message from 'module' to default…

d0a026e

… value 'function'Removed setting of message.chat.username to None in test 'test_build_reply_arguments_chat_id_and_message_id' since it is no longer needed.executed multiple times 'pytest -m no_req -n auto --dist=loadgroup --lf' to validate all tests in test_message.py will still pass.