Ton of failures along the lines of:

======================================================================
ERROR: test_ctor (test.test_asyncio.test_unix_events.UnixReadPipeTransportTests)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/Users/bcannon/Repositories/cpython/default/Lib/test/test_asyncio/test_unix_events.py", line 329, in setUp
    self.loop = self.new_test_loop()
  File "/Users/bcannon/Repositories/cpython/default/Lib/asyncio/test_utils.py", line 413, in new_test_loop
    self.set_event_loop(loop)
  File "/Users/bcannon/Repositories/cpython/default/Lib/asyncio/test_utils.py", line 407, in set_event_loop
    events.set_event_loop(None)
  File "/Users/bcannon/Repositories/cpython/default/Lib/asyncio/events.py", line 581, in set_event_loop
    get_event_loop_policy().set_event_loop(loop)
AttributeError: 'object' object has no attribute 'set_event_loop'

-O disables assertions, and asyncio has a ton of assertions. Some tests maybe rely on them? I guess that they are real bugs (assert must be replaced with a regular if+raise) in the middle of the failures.

This issue is still relevant 4 years later: many test_asyncio tests fail with python3 -O :-(

I'm sure there's more to it, but I found at least this failure:

~/cpython$ ./python.exe -O -m test test_asyncio -m test_set_event_loop
Raised RLIMIT_NOFILE: 256 -> 1024
0:00:00 load avg: 2.02 Run tests sequentially
0:00:00 load avg: 2.02 [1/1] test_asyncio
test test_asyncio failed -- Traceback (most recent call last):
  File "/Users/guido/cpython/Lib/test/test_asyncio/test_events.py", line 2595, in test_set_event_loop
    self.assertRaises(AssertionError, policy.set_event_loop, object())
    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
AssertionError: AssertionError not raised by set_event_loop

test_asyncio failed (1 failure)

== Tests result: FAILURE ==

1 test failed:
    test_asyncio

Total duration: 134 ms
Tests result: FAILURE


Also this one:

~/cpython$ ./python.exe -O -m test test_asyncio -m test_create_datagram_endpoint_addr_error
0:00:00 load avg: 2.19 Run tests sequentially
0:00:00 load avg: 2.19 [1/1] test_asyncio
test test_asyncio failed -- Traceback (most recent call last):
  File "/Users/guido/cpython/Lib/test/test_asyncio/test_base_events.py", line 1593, in test_create_datagram_endpoint_addr_error
    self.assertRaises(
    ^^^^^^^^^^^^^^^^^^
  File "/Users/guido/cpython/Lib/unittest/case.py", line 734, in assertRaises
    return context.handle('assertRaises', args, kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/guido/cpython/Lib/unittest/case.py", line 218, in handle
    callable_obj(*args, **kwargs)
    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/guido/cpython/Lib/asyncio/base_events.py", line 637, in run_until_complete
    return future.result()
           ^^^^^^^^^^^^^^^
  File "/Users/guido/cpython/Lib/asyncio/base_events.py", line 1287, in create_datagram_endpoint
    infos = await self._ensure_resolved(
            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/guido/cpython/Lib/asyncio/base_events.py", line 1369, in _ensure_resolved
    info = _ipaddr_info(host, port, family, type, proto, *address[2:])
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
TypeError: _ipaddr_info() takes from 5 to 7 positional arguments but 12 were given

test_asyncio failed (1 error)

== Tests result: FAILURE ==

1 test failed:
    test_asyncio

Total duration: 165 ms
Tests result: FAILURE


I think calling assertRaises(AssertionError, ...) is definitely an anti-pattern. We can't remove all assertions from the asyncio library, but I think the ones that are explicitly checked for by the tests should go, at least. (Probably replaced with TypeError or ValueError in most cases.)

I'd be happy to mentor someone who wants to tackle this.

I would like to tackle this, I'll work on this from next week.

New changeset 265918bb1d782ab85c7dbc835eb62d6cfc2145b7 by Kumar Aditya in branch 'main':
bpo-23819: asyncio: Replace AssertionError with TypeError where it makes sense (GH-29894)
https://github.com/python/cpython/commit/265918bb1d782ab85c7dbc835eb62d6cfc2145b7

New changeset 6ca78affc8023bc5023189d64d8050857662042a by Serhiy Storchaka in branch 'main':
bpo-23819: Get rid of assert statements in test_asyncio (GH-30212)
https://github.com/python/cpython/commit/6ca78affc8023bc5023189d64d8050857662042a

New changeset 95948169d75bed3936284ea2225e83e07ec5fe20 by Miss Islington (bot) in branch '3.10':
bpo-23819: Get rid of assert statements in test_asyncio (GH-30212) (GH-30213)
https://github.com/python/cpython/commit/95948169d75bed3936284ea2225e83e07ec5fe20

New changeset a23ab7b6d8b3ae3a47747c0c4bceb2370cc48dcc by Kumar Aditya in branch 'main':
bpo-23819: Fix asyncio tests on python optimized mode (GH-30195)
https://github.com/python/cpython/commit/a23ab7b6d8b3ae3a47747c0c4bceb2370cc48dcc

New changeset 07229054a129a72b4ffdf29252eb73c6154c0ccf by Miss Islington (bot) in branch '3.9':
[3.9] bpo-23819: Fix asyncio tests on python optimized mode (GH-30195) (GH-30265)
https://github.com/python/cpython/commit/07229054a129a72b4ffdf29252eb73c6154c0ccf

Is anything left to do?