In article <34F1CCEF.98AA0710@undergrad.math.uwaterloo.ca> you write:
> RAM and large amounts of disk space.  However, they can't really handle
> the binary bloat because the CPU performance and disk data transfer
> rates aren't that great.  No matter what you do, it *will* significantly
> slow down the system.  Right now you can more or less get away with it

Not necessarily. Often the bloat is the result of a time-space tradeoff.
Consider that NetBSD useds GCC, which is happy to generate larger code
in exchange for increased execution speed. Likewise, changes in
executable file format to make sure each section begins in a new page
create larger files, but may allow more efficient paging, etc.

> things would be more pleasant without the bloat.  As for running a real
> multiuser system, I'm sure that 4.3BSD would perform better than NetBSD
> on an old VAX just because the binaries are smaller.  

That's not a good assumption.  It's quite possible it's true.  It's also
quite possible it's not.  I'd sure want to benchmark before I did
something as drastic as going to 4.3BSD, considering all the potential
pitfalls (security, etc...  you-know-who's 4.3BSD network is going to be
a cracker haven, if it ever happens).  And implementing shared libraries
is a saner solution.  Plus, NetBSD has performance improvements in some
areas that may make up for losses in others, and will gain even more
over time. Anyone looked at supporting UVM on NetBSD/vax?