Huh, that's weird. My first thought was some kind of race condition, but... raise_signal uses raise(), which on Windows should be invoking the signal handler synchronously, so the warning should definitely be printed before raise_signal() returns.

Could the warning be trapped in some buffer? That would be weird too, usually stderr and warnings should not be buffered.

Of course putting a retry loop around the test is an option if we can't figure out how to fix it properly.