Povilas Kanapickas

Sorry, didn't notice your reply.

@glyph I did some investigation. pytest is to blame. See their code that is patching twisted Failure __init__ here: https://github.com/pytest-dev/pytest/blob/main/src/_pytest/unittest.py#L421. The difference between implementations is that due to async/await we...

@voutilad I think that this PR can be closed. The fixes have been independently landed in https://github.com/scylladb/seastar/pull/2535. Unfortunately I didn't think to look into the PR list before spending time...

> Why is that a problem? The way it's described here it looks like -- a code sends two messages into output stream, and those two messages are squashed together...

I think it makes sense to consider waiting at least a year after Debian 13 is released.

Fixed in https://github.com/buildbot/buildbot/pull/7895 and backported to 4.0.3.

@tdesveaux > I'll take a stab at patching make_master to make it await it (and everywhere it's called...). Sorry for noticing this only now - I have the fix already...

``` Error: Traceback (most recent call last): File "C:\hostedtoolcache\windows\Python\3.13.0\x64\Lib\site-packages\twisted\trial\_synctest.py", line 1071, in run warnings.warn_explicit(**w) builtins.ResourceWarning: unclosed database in buildbot.test.unit.changes.test_hgpoller.TestHgPollerBranches.test_poll_regular ``` Amazing. This warning will allow us to track a number...

I think this feature makes sense in principle, but instead of environment variable and command-line option I would add a customization point to the compose file. For example, we already...

A couple of nitpicks: > The existing io.podman labels are already "undefined behavior"/"out of spec" (not in itself necessarily an issue!) This is a conscious decision due to the fact...