Adam Williamson

> FICS usually has several bots online to play with. Thanks, I found out how to identify them - (C) in the name. So I was able to test my...

that sounds like you didn't actually get the fix - if you did, at least the error message should be slightly different.

I did a scratch build of 1.0.5 with the other fixes plus this one, for Fedora 42, at https://koji.fedoraproject.org/koji/taskinfo?taskID=135153953 - can you test that?

oh, no, I see the same: ``` File "/usr/lib/python3.14/site-packages/pychess/ic/TimeSeal.py", line 67, in data_received cooked = self.cook_some(data) File "/usr/lib/python3.14/site-packages/pychess/ic/TimeSeal.py", line 93, in cook_some self._stream_reader.writer.write( ^^^^^^^^^^^^^^^^^^^^^^^^^^ AttributeError: 'ICSStreamReader' object has no attribute...

yeah, no, the fix is just broken AFAICT. `ICSStreamReader` inherits from `asyncio.StreamReader` and neither of those have a `writer` property. `asyncio.StreamReader` does have a `_w*a*iter` property (emphasis added) - maybe...

oh, this also incorporates a fix I had to do to the *upstream* one to make it work with openssl 3.2 - https://github.com/python/cpython/pull/120764

on a quick look, the failures look like flakes, not to do with this PR...

If you care about keeping the tests running on ancient openssl, it would also be good to check whether they still do. It's actually hard for me to do that...

Hmm, reviewing the PR, we do lose the change to make the certs valid after 2037. It should be easy to change that by changing the `enddate` variable, I can...

Well, I suspect it might have been done so you can test 2038 problems by testing with the system date set past 2038. I guess I can kick it out...