At some point, we'll need to shift gears from fun feature development
and focus on brutal stress testing. I still have a sigh of relief
after any nontrivial systemtap script runs to completion; I still
don't dare run much stuff on my workstation. What about you?
It is still way too easy to crash a machine, and I just don't see
enough being done to toughen the system. What ever happened to Brad
Chen's collected test plan? It might already be too late for the next
RHEL4 update for stability improvements to (say) kprobes to squeeze
in, but what other deadline should we choose as a refocusing trigger?