Setting up Domains

[miller@NBS-ONE.cam.nist.gov (Bruce R. Miller)]

We have only a few lispms 3-4, but we've got a lot of suns, irises, etc.
Based on this, or just my own prejudices, I didn't want to establish a
whole new level of domain for our lispms; I want them in the same
domain as the rest of our center (CAM.NIST.GOV).

So, I decided not to have any lispm be a domain server, I would rely 
on a sun (FS1, the ONLY host in namespace providing :domain service)
to do that.  I also have to maintain certain parallel
info in the namespace. But since I'll only keep the lispms and other
`important' hosts there, file servers, etc, that shouldn't be too bad; I thought!

However, it turns out there is another problem.  
If I dont have the internet-address in our namespace, then
"gem" is a lispm-host, but gem.cam.nist.gov is a RANDOM-host.
On the otherhand, if I DO give the internet address in our namespace,
then some foo.cam.nist.gov might be found in the namespace, or it might be
found by asking FS1.

Unfortunately, code like (net:parse-host) decides that if it CAN be in 
the namespace, then it MUST be, and wont ask FS1!

I haven't found a clue in the doc as to whether this is INTENDED to be do-able or not.
While I was poking in sys:network;domains looking for possible missed namespace 
entries, or whatever, trying to find out where things failed, I came up with the following:

If I patch the call to 
      NETI:FIND-NAMESPACE-DOMAIN  within NETI:FIND-DOMAIN
so that it NEVER :create's an unknown object, I get what I (think I) want.
Namely, anything foo.cam.nist.gov NOT in the namespace AINT SUPPOSED to be, and
should be found by NETI:FIND-NATIVE-DOMAIN, ie. using :domain service.

So, the questions are:
  Is there a `proper' way to do this?
  Should I really just create a sub-domain? Why?
  If I do it this way am I also getting something I'm not aware of (and dont want)?
  Is there a better way?
  I think this all applies only to Hosts; If I'm wrong, that could mess me up too.
  Is this in any way sane?
  ...
bruce
miller@cam.nist.gov

(well, that works at least!)