Drop 3.9, Start testing on 3.14 and 3.14-free-threaded, get tests passing again #1076
Conversation
Carreau
changed the title
Carreau
force-pushed
the
314
branch
4 times, most recently
from
e57ecba to
338dd19
Compare
|
I do not know how to handle or debug the remaining failures. |
|
@Carreau want me to take this one over? I can have a stab at it. |
|
Sure, and feel free to push on my branch. I'm debugging some issues on pytest-jupyter as well. It might be due to the Event Loop policies deprecation warnings that are the source of the issue. |
not really for runtime, but needed for compatibility with recent pytest
hatch gets in the way a lot more than it helps - stop re-running failing tests - show the test environment
|
There are definitely once again sockets and contexts being created and not closed in the tests, and we need to track those down. |
it works fine locally
sometimes it just ignores it (Free threaded)
|
should all those non-closed sockets and open/close pairs become context manager (in a perfect world). |
Yes, KernelManager and KernelClient should both be context managers. zmq sockets are already context managers and have been for some time |
don't override hatch doesn't work on windows, let it do the wrong thing
improves cleanup
startup sequence seems really slow now
minrk
changed the title
|
Restarted MacOS test, failure was intermittent. |
Carreau
left a comment
Carreau
left a comment
There was a problem hiding this comment.
+1, a couple of questions, but good as is.
| hatch run cov:test --cov-fail-under 75 || hatch run test:test --lf | ||
| - name: Run the tests on pypy | ||
| echo "PYTEST_ADDOPTS=${PYTEST_ADDOPTS:-} -W default" >> $GITHUB_ENV | ||
There was a problem hiding this comment.
Do you want this to be permanent ?
There was a problem hiding this comment.
Yes, this is not a change, it simplifies the 3 different test invocations below by setting the options before using the same pytest invocation.
Co-authored-by: M Bussonnier <[email protected]>
|
green again. One flaky test, still, but passed on re-run. |
2 participants
No description provided.