IPP Mail Archive: IPP> What would the world be like if the host to printer requirements

IPP> What would the world be like if the host to printer requirements

Paul Moore (paulmo@microsoft.com)
Thu, 12 Feb 1998 11:45:46 -0800

I apologize for not attending the teleconference - the minutes make
interesting reading.

If we decided that there would a separate group doing 'not LPD and SNMP'
(NLS) then several things would happen in the IPP world:-

1. The notification debate would be greatly simplified. IPP does client
notification in human readable form, the protocol itself does not carry the
notification, merely the request that the receiver generates the
notification, eg attributes notify-method=email, notify-events=all,
notify-destination='paulmo@microsoft.com'. There would need to be some way
that a client could ask a server what type of notifications it supports.

2. IPP would be liberated from the requirements that this stuff fit in a
printer's NIC.

3. IPP can focus on adding value between the client and server (enumerate
printers, redirect jobs, etc.)

The NLS protocol then picks up the requirements for:-

1. Low level device discovery - a server hunting around for a new device. (A
server is not going to do a Yahoo search!)

2. Discovery of device capabilities (in general, 'what can a PrinterRUs mega
laser 40 do?' plus 'specifically what can this megalaser 40 do?').
Discovering how to exploit the published feature set.

3. The peer queuing issue gets addressed cleanly.

4. Notifications get sent asynchronously , in machine readable form to well
defined sets of end points (subscriber model).

5. We can talk about transport neutral again.

6. We can have good, defined security. (I would do this by certificate
passing, with the printer having a list of certificate issuing authorities
it trusts - 'any friend of verisign's is a friend of mine').

7. We dont care about tunneling though firewalls, etc. - that is IPP's job

8. managing device settings, resetting the printer, .....

I dont think software builders would have any problems with this model, we
already have a two step internal model, app to print subsystem, print
subsystem to hardware marking device. Hardware builders have a real problem
- Which protocol do they put in a printer? Or maybe both. That in turn
creates a problem for the s/w people because we need the h/w people to
support NLS or else it is no use.