Re: patch linus

Prend cet e-mail par example: https://lkml.org/lkml/2012/12/23/75

Voilà comment je l’aurais écrit:

this may very well be a bug in pulse audio. We unfortunately do not control pulse audio and need to work around it. Remember the golden rule: we do not break userspace.
It is what it is, it’s unfortunate, but we have no choice but to put up with it, at the very least until the PA guys fix it.

commit f0ed2ce840b3 is also technically wrong. ENOENT is not a valid error return from an ioctl. ENOENT means "No such file and directory", and is for path operations. ioctl's are done on files that have already been opened, making it invalid to return ENOENT from an ioctl. This particular point isn’t open for discussion. We need to fix this immediately.

I have another error report in my mailbox of all KDE media applications being broken by v3.8-rc1, and suspect it’s the same kernel bug. Given the urgency of this matter, I’ll handle it personally.

Et potentiellement, en privé, pas sur la ml

I’m quite frustrated with your latest patch. I want to reiterate and stress out that it is critical that the kernel does not break user space code. We’ve always followed this rule.
It doesn’t matter how broken the user space code is, it has shipped and is used in production. We need to be very careful about this.
I’d suggest you be a bit more careful going forward. There is a big code smell, where the patch is returning an invalid code, and then attempting to correct it in a few places. This is a clear sign that the return code was wrong in the first place.

J’ai pas la moindre idée de ce que son compliance tool fait, mais ya moyen d’aborder la discussion la dessus différemment aussi.

Bref, tu peux faire passer exactement le même message diplomatiquement et sans traîner un contributeur dans la boue et en public.