07:13 <mattip> py3.8 failed the test_fork_hook_creates_thread_bug in extra_tests/test_posix.py

07:13 <mattip> that sounds like something that might influence compileall and multiprocessing

08:02 <ctismer> mattip: that alpha has 46 bugs instead of 25. Something is broken.

08:03 <ctismer> (in our PySide tests)

08:06 <mattip> which alpha?

08:07 <mattip> can you try with a nightly?

08:23 <ctismer> The same nightly from yesterday where I found the setuptools bug

08:43 <mattip> hmm. Is there any pattern to the new failures?

09:42 <ctismer> well, I have a script that runs through all objects and generates .pyi files. That runs automatically in the build. Now it segfaults.

10:28 <mattip> with a stacktrace?

12:47 <ctismer> hey mattip sorry my internet is flaky, today, only cellphone. Here we go:

12:47 <ctismer> mattip: https://www.irccloud.com/pastebin/X3ndyANu/pypy%203.8%20alpha0

12:49 <ctismer> (sorry, no segfault, just crashing)

12:51 <ctismer> When I disable that script, then I get 46 errors instead of 25.

12:54 <mattip> that error should have been fixed on the nightlies

13:00 <ctismer> ok, I'll check again :)

13:17 <ctismer> building...

13:41 <ctismer> mattip: here the current alpha0

13:41 <ctismer> mattip: https://www.irccloud.com/pastebin/ed8bm73J/generate_pyi

13:44 <ctismer> Which is this line:

13:44 <ctismer> # find out how many functions create a signature

13:44 <ctismer> sigs = list(_ for _ in functions if get_sig(_[1]))

13:58 <mattip> checking ...

14:00 <mattip> it would be nice if pypy would print out the function name in the error message

14:00 <mattip> SystemError: Function returned a NULL result without setting an exception

14:03 <mattip> ctismer: do you have any idea what are the arguments to "generate_pyi(import_name, outpath, options)"

14:03 <mattip> when it crashes

14:06 <ctismer> mattip: I can check tomorrow. Today we have guests and no DSL, Telecom brojen

14:06 <ctismer> Broken

14:06 <mattip> thanks

14:06 <mattip> I am off for a break too