[issue24397] Test asynchat makes wrong assumptions

Change by Irit Katriel : -- resolution: -> wont fix stage: -> resolved status: open -> closed superseder: -> Close asyncore/asynchat/smtpd issues and list them here ___ Python tracker

[issue24397] Test asynchat makes wrong assumptions

Change by Irit Katriel : -- versions: +Python 3.10, Python 3.8, Python 3.9 -Python 2.7, Python 3.2, Python 3.3, Python 3.4, Python 3.5 ___ Python tracker ___ _

[issue24397] Test asynchat makes wrong assumptions

Étienne Buira added the comment: That would just race the other way around. However, I missed the fact that the check of client's contents had no chance to catch an issue, as the main thread were not entering asyncore.loop after allowing the server to send. Updated patch (against 3.3) follows

[issue24397] Test asynchat makes wrong assumptions

Changes by Étienne Buira : Removed file: http://bugs.python.org/file39643/test_asynchat_check_received_len_if_received_len_matters.diff ___ Python tracker ___ __

[issue24397] Test asynchat makes wrong assumptions

R. David Murray added the comment: The comments in the test indicate that this possibility was considered and dealt with. So that indicates it is a test bug in this particular test (test_close_when_done, for other readers), not in the fixture. It looks like the race condition still exists, t

[issue24397] Test asynchat makes wrong assumptions

Étienne Buira added the comment: The issue is that this buffer is also emptied at the end. Current version assumes that the server thread will not have enough opportunity to send the whole buffer before receiving a RST. But in reality, it's quite possible (happens reliably on a box) that: 1. m

[issue24397] Test asynchat makes wrong assumptions

R. David Murray added the comment: (Calling that variable 'buffer' is a bit misleading...it isn't really a buffer that gets filled, it is an immutable bytes object that gets replaced by a new one with the additional received data appended.) -- ___ P

[issue24397] Test asynchat makes wrong assumptions

R. David Murray added the comment: Your emendation doesn't seem to do anything differently (logic wise) than the original code. Unless I am missing something, it is pointless to track the length of the buffer separately from the contents of the buffer. The buffer knows its length. -

[issue24397] Test asynchat makes wrong assumptions

New submission from Étienne Buira: Amount of data the server could not send back is not a reliable indication on how much data it received. -- components: Tests files: test_asynchat_check_received_len_if_received_len_matters.diff keywords: patch messages: 244922 nosy: eacb priority: nor