I would strongly suggest that NEW_BUFQ_STRATEGY *not* be made the default. I
just experienced a problem that I had experienced the first time I tried it
out, close to a year ago:

I'm running in X, with opera running, a few tcl/tk apps, and other things,
and compiling a very bloated c++ program. The machine lagged so hard that I
thought it was deadlocked for about 30 seconds, then it was usable again,
then it did it once more for about 10 seconds, then went back to usability.
This has never once happened without NEW_BUFQ_STRATEGY. I only re-added
NEW_BUFQ_STRATEGY to my kernel because of the claims that it had stabilized
- I think that is not the case.

This is NetBSD/i386 1.6ZF.

-Sean

--
/~\ The ASCII
\ / Ribbon Campaign                   Sean Davis
 X  Against HTML                       aka dive
/ \ Email!