I like that Linus is so strict on not breaking user space because this obviously aids with compatibility and it’s probably a big part of why rolling releases work.
I think kernel still has compatibility with paleolithic glibc enabled by default
ENOENT is not a valid error return from an ioctl. Never has been, never will be. ENOENT means “No such file and directory”, and is for path operations. ioctl’s are done on files that have already been opened, there’s no way in hell that ENOENT would ever be valid.
Yes. I did not include patch from first person in screenshot because I thought it would make it too boring to read. But it kinda adds even more to context.
Replying to “I get this regression with KDE on this system caused by this commit and here how I fixed it” with “lol, pulseaudio sucks, opensuse sucks” of course will make Linus angry and he will reply not only “no u”, but also “and here’s why”.
So Linus should have sided with someone who in regression report of KDE using 100% CPU starts blaming pulseaudio and opensuse and double down on blaming pulseaudio? Instead of fixing syscall returning completely unrelated error code. It’s like if your router crashed with message “there is no milk in your fridge”.
Linus does not fuck around (lemmy.one)
An oldie, but a goodie
What the hell! Let's all just go crazy! (lemmy.world)