🚀 go-pugleaf

In <9224017.23439@mulga.cs.mu.OZ.AU> kre@cs.mu.oz.au (Robert Elz) writes:

>In <1992Aug26.143358.28814@minyos.xx.rmit.oz.au> s892024@minyos.xx.rmit.oz.au (Richard A. Muirden [GA]) writes:

>| OK here's my 2 cents, why not combine the two and have:

>| 	node.state.priv.au
>| 	eg:
>| 	node.vic.priv.au

>Because I'd prefer not to have anything geographic in the name at all,
>or not for a private system that is quite likely at some time to decide
>to relocate to some other state.   Not because anything would need to alter
>if that happened, but because people would always wonder why the system
>belonging to someone living in Melbourne is called bloggs.wa.priv.au
>or something - there'd be a lot of pressure (and perhaps some people
>thinking it was necessary) to change the name to bloggs.vic.priv.au
>which is the very thing that shouldn't be necessary.

Here's another suggestion, similar to C. Vance's.

	surnametag.ident.priv.au

where "surnametag" is derived from the user's surname, say with the first
7 letters and a digit.  e.g. "elz1", "weather2", etc.  "ident" is some uniquely
AARNet-allocated identifier to split the namespace up better.  What are these
identifiers to look like?  Search me.  Derived from the current date?

There just needs to be some way to allocate names with some degree of
non-overlap, while allowing for other "elz1"'s and "weather2"'s in the future
under a different identifier.  If the identifier was derived from the date
say (which is not really something I like personally - see below), then if
two people registered in the same month with the surname prefix "xyzabc",
they'd become "xyzabc1" and "xyzabc2" under whatever the current identifer is.
This will keep the burden light on the mug ... er ... individual ( :-) )
handing out the names so that it is only necessary to keep track of the names
allocated in the last month.

Machine names would have one of the following forms:

	surnametag.ident.priv.au
	machine.surnametag.ident.priv.au

The first for a single-machine person, and the second for hardware-aholics.
I think the first would read very well in e-mail addresses.  For example,
"rhys@weather2.abc.priv.au".

Why don't I like date-derived identifiers?  Then people could point at the
host name and say "you've only been on the net 3 years, you don't know
anything", which is not something I'd like to see happen.  Depends on
how the derivation is done I suppose.  Start at 1 and add one each month?

On the other side, there _is_ something that can be said about organisations:
there is only one MX for the AARNet service provider to worry feeding to a
particular private network instead of 100's.

One more thing before I shut up: will a "priv.au" MX registration come with
a $1,000 AARNet-usage fee?  Very few private individuals can wear that kind of
cost (together with the service provider costs), especially when they don't
generate as much traffic as the universities, corporations, etc.  One of the
motivations behind forming BrisNet was to share the costs around so that it
becomes possible to get news and mail, even on a shoestring budget.  4Mb a
month is an awful lot of mail traffic for one person.  There are lots of
people out there itching to "get on", but get discouraged at the costs.

Cheers,

Rhys.
--
Rhys Weatherley, University of Queensland, Australia.
rhys@cs.uq.oz.au  "I'm a FAQ nut - what's your problem?"