Re: Using qsort(3) in the kernel (was: CVS commit: src)

[Andrew Doran <ad%netbsd.org@localhost>]

> > Our implementation of qsort() uses recursion. That is IMHO a dangerous
> > thing to do in the kernel.
> 
> Another good reason to prefer mergesort is that it is a real time
> algorithm -- no degenerated cases.

This isn't directed at anyone in particular: while this will likely lead to
an animated discussion, it doesn't matter much what we provide as long as it
works, because if you're sorting large chunks of data in a critical kernel
path you have already lost.

Cheers,
Andrew