It's usually a tough position to be in when one complains about a free
service or product, but I do understand Peter's feelings here.  I've got
a 5k/133, a 5k/260, and a pile of 5k/25's here begging to be used for
something.  I'm sure they'd be just dandy for *something*, like acting
as DNS resolvers, but UBD (Ultrix Brain Damage) makes it tough for me to
put them back into service, especially with the silly 2-user
enforcement.  For the sorts of uses I could make of these boxes, I don't
even need an X server or client libraries.  What I need most is a
working system that I can install, boot, and run -- from a serial
console.  I probably need to be able to build a working kernel, too.  If
I could have just this, I coulld bring a handful of these boxes into
service and probably contribute back various utility ports.  I regret
that I don't have the time or background to help perfect the kernel or
installation procedures.

Yes, I've read through the last handful of months of list archives, but
have found confusing and sometimes contradictory statements.

Don't get me wrong -- I appreciate the efforts that have been made, and
am somewhat in awe that people can port a kernel to a new architecture
at all in their free time.  Two things that I'd like to see for netbsd
as a whole:

1) A pan-architecture statement about why it's being ported to so many
platforms in the first place.  In some cases, like PMAX, it's clear --
UBD.  In others, like SPARC, it's not so clear.  Certainly the idea of
having an identical OS across disparate platforms is a valid one, though
I wonder just how identical it is.  Claims are made that it allows one
to run a "modern" OS instead of an "unsupported" one, though I wonder if
"modern" really means "nothing compiles on it".  I've got a number of
boxes here running BSDI, which is "supported", yet it's often
frustrating when things don't build or work correctly on them, or when I
can't find answers to questions that can easily be answered for more
common platforms -- such as the relationship of vnodes to the namei
cache, and how I properly tweak whatever I need to to to the equivalent
of enlarging it so that news pathnames stay cached for performance wins.
 I can easily find this information for both recent flavors of SunOS,
for example, but have been frustrated in finding it for BSDI, which is
probably similar to *BSD in this respect.  I've also been led to believe
that NetBSD doesn't do shared libraries.

2) References are made to SUP'ing source trees, yet nowhere can I find a
description of what this means.  The fact that the pmax and mainstream
netbsd trees seem to be divergent doesn't help me figure out where to
jump in.


Again, I don't want to come off as flaming people when I haven't
contributed a damned thing yet to the effort, but felt the need to
express the things that are keeping me from joining the club.