Philip

APC has a good lengthy interview with Con Kolivas about why he’s decided to quit from Linux kernel development after several setbacks in his work with the swap prefetch, [Deadline] Staircase Scheduler and more. Kind of sad really, from what I could tell he developed a user-oriented interactivity based CPU scheduler that helped the desktop GNU/Linux users. It was then rejected and rewritten by other kernel developers in what appears to be a huge case of NIH (Not Invented Here) syndrome citing lack of objective benchmarks as being the reason why.

Read it here.

People who know me probably know I don’t like the Linux way of doing things at times, this is a pretty good example. Con developed a pluggable scheduler system that would have let users choose which one to use, yet Linus and other developers keep insisting that a “one size fits all” scheduler is all that is needed. Monolithic design, oh yeah…