(This is not a bug report; it is a request for comments)

I don't know if this exists beyond the i386 port, but that's where
I observed it.

I recently got my notebook running NetBSD-1.0 on the network at work
thanks to Stefan Grefen's PCMCIA work (Hi Stefan!).  We run NIS here, 
so the first thing was to get helios slurping NIS from our servers.

That was more than easy enough.  Everything was dirt bag fast:
telnet, rlogin, ftp, etc...

Our site is going to be converting from NIS to DNS for host information
soon.  So, I have a shell script that basically does a ypcat on hosts and
feeds it to O'Reilly's h2n perl script.  It will still need some cleaning,
but it's a good place to start.

After I have name service in place, something odd happens.  rcp, rsh, rlogin
are all much slower.  Sometimes, a request sent for a xterm for example
never appears, but after you login via telnet, all the xterms requested
suddenly appear on the display specified.

telnet and ftp are still very fast.  It seems to be the family of r* commands
that have this 'delay' problem.

The order of items in resolv.conf do not make a difference in the speed
of the r* commands, so I'm at a loss.  Has anyone else experienced this
kind of slow down when using DNS vs NIS on NetBSD-1.0?

BTW, this is an unpatched stock NetBSD-1.0, I currently don't have the
space for the source tree just yet.  Is this a problem that perhaps one
of the patches resolves?