test_asyncio failed: flaky test_sendfile_close_peer_in_the_middle_of_receiving

[tasty0tomato]

I built 3.11.5 on my own Windows computer, and found it passed this test on one computer, but didn't pass on another computer. I think that this issue may not be fixed completely, or I can provide needed information of the computer which this test didn't pass to find out the reason why it didn't pass.

(Originally posted in #85848 (comment))

[sobolevn]

Please, provide more information: os name and version, test output, how python was built and executed.

[tasty0tomato]

OS name: Windows 10

OS version: 19045.3516

Test output(only test_asyncio):

0:00:44 [ 26/434] test_asyncio
test test_asyncio failed -- Traceback (most recent call last):
  File "C:\Python-3.11.5\Lib\test\test_asyncio\test_sendfile.py", line 473, in test_sendfile_close_peer_in_the_middle_of_receiving
    self.assertTrue(1024 <= self.file.tell() < len(self.DATA),
AssertionError: False is not true : 0

How python was built:
Just followed PCbuild/readme.txt, ran build.bat.

[sobolevn]

It has a big history: https://github.com/search?q=repo%3Apython%2Fcpython+test_sendfile_close_peer_in_the_middle_of_receiving&type=issues

Thanks for the details, this test is known to be problematic. You can just ignore it for now.
It should be fixed on our side :)

[tasty0tomato]

There is one more interesting information. This test passes sucessfully on my another Windows 10 19045.3516 computer. So I think maybe OS is not the reason.

[gvanrossum]

Wasn't this fixed?

[tasty0tomato]

Yes, at least this problem is fixed in 3.11.7, but I find another issue in 3.11.7 which has been reported in #113026.

[tasty0tomato]

Well, because of some reasons, I need to build a python3.8.19. And it also has this kind of errors. But it seems that python 3.8 will be eol by the end of this year. So, anyway.