A kernel change breaks GlusterFS

Bernd,

It was not my intention to assign blame, and I'm sorry if the article gave that impression. The change that Fan Yong (or something like it) made was obviously needed. And, at some level, I could imagine that the FS developers didn't see this as an ABI change. My point was a general one: even things that may not look like ABI changes may in fact be so, and developers as a group perhaps need to be even more vigilant about that possibility. (The message queue change is a similar kind of case. One could easily have thought of it as *not* being an ABI change, but it was.)

Thanks,

Michael