> He benchmarked NetBSD-current yesterday. 
> http://bulk.fefe.de/scalability/#newdata
> Some benchmarks declined a bit, but the last and IMHO most important 
> 'http-request-latency' benchmark comes close to Linux/FreeBSD. 
> (1.6/poll vs. 2.0/kqueue)
> 
> And he encountered a problem during the update:
> "The upgrade build process did not run smoothly, the NetBSD crowd will 
> have to review that (e.g. I got a /usr/src/usr.bin/kqueue/kqueue/ 
> directory from CVS, which the make clean run in build.sh tried to rm -f 
> (without -r), and that caused the whole build to abort)."
> 
> He didn't wrote when he checked-out the source, but it must be between 
> 19. and 20. Oct.

	he did not specify "-P" on cvs update, i guess.

itojun