I ran FreeBSD for the past couple of years, but I recently switched to a
"fixed" NetBSD 1.0 installation.  I did so because of the incredible bloat
that the FreeBSD kernel has acquired.  For example, one NetBSD 1.0 kernel I
build comes out below 500K, while the equivalent FreeBSD 2.1 kernel sprawls
over 700K (where "equivalent" means as close as possible in options and
devices).

FreeBSD has degenerated into a wanna-be SVR4.  Yet during the same period
SVR4 has been left to wander, so to speak, and has finally wound up on SCO's
desk, all without anyone paying much attention.  The FreeBSD project might
well ask themselves why that happened, and so might you.  Once upon a time,
AT&T called SVR4 "Destiny."  Some destiny.

Now in assessing NetBSD 1.1_ALPHA, I notice that applying the same build
results in a net gain of about 35K in the kernel size, above 1.0.  The only
item I can find in the CHANGES file to explain this is something about
"filling in the missing kvm code."  Yet I see no functional changes of note,
other than the build size.

So I log into this news group, and see mostly this thread on what to do with
the VM system, including suggestions to adopt FreeBSD's code.  While I agree
there is a problem worthy of attention, I worry that the proposed cure might
be worse than the disease.


--Jim Howard