sorry anselm: i see that the original message was repeated just now.
anyway, it's still true that we went over this one during the last few days.

the short answer is: the cpu times are basically the same on
both Plan 9 and the Other System except when doing unbuffered writes (ie, print)
on Plan 9 (and that to a non-local file, not /dev/null or ramfs).
at least, i get exactly that effect on the Plan 9 and Linux systems i was timing,
given the bug-fixed source code that was the topic of discussion.