On 5/1/22 13:28, Bruno Haible wrote:
I would argue that glibc should use a different errno value in this case.
Either errno value makes sense to me. If you keep doing a non-blocking read on a pipe that only you can write to you'll keep getting EAGAIN, which has the same feel as a bug where you grab a lock and then keep doing a malloc that won't work until you release the lock.
No big deal either way of course.
