🚀 go-pugleaf

In <5w43wB5w165w@willard.atl.ga.us> dawson@willard.atl.ga.us (Willard Dawson) writes:
>djk@boombox.apana.org.au (David Keegel) writes:

>> If I want to send mail to a particular machine, I don't want to have
>> to care whether it connects to the world with UUCP or TCP/IP or avian
>> carrier.  And I certainly don't want to change my address book because
>> they decided to use SLIP instead of UUCP as a transport to their link.

>Granted, that might make sense.  (One possible exception to that is, if
>it were possible for you to specify the delivery mode [airmail? overland?
>each are usually options for the traditional mail, as per your own example])

Specifying delivery modes in a meaningful way would be very difficult
at present with the number of different networks which e-mail can go
though which have different characteristics and "standards".

While UUCP has grades (and perhaps even alternate routes, manually),
and ACSnet has independant "speed" and "cost" metrics (which I think
is what you really had in mind), talking about modes on the Internet
makes about as much sense as modes when sending faxes.  And who knows
what the myriad other protocols (fidonet, osi, X.25, ...) do.

Translating between all of these would be a real nightmare.
If you really want to do this, probably the best way would be to have
intelligent gateways between networks translate Precedence: headers
into something appropriate for that network.

>> If UUCP is so different to every other transport layer in the world
>> that it causes sites which use it to be a separate "networking system"
>> that cannot sensibly correspond with other networking systems, then I
>> don't think it's doing a very good job.

>It's not that UUCP is doing things differently, it's that inconsistency
>amongst Internet sites (some of whom treat .UUCP as a "valid" domain, by
>routing mail to such sites, and some who do not) that causes me to be
>concerned.

The point I was trying to make (or thought I was trying to make) was that
there's a lot more to E-mail than just The Internet and UUCP, and if you
want to be a real part of the worldwide e-mail network, you have to co-
operate a little with existing practices; trade off your autonomy to be
part of a larger entity.

One way of looking at this is that The Internet tells you what to do and
you have to do it.  Another way is to see that there are smart guys who
have figured out (part of) what it takes to have a smoothly running (:-)
e-mail system, and have already forced that model on The Internet.

As a for instance, such a system does *not* include have a flat naming
space for anything but trivial networks.  The people who hang around on
The Internet have already found this out.  Learn from their mistakes.

It is up to you whether you see them as good Samaritans or Jehovah's
witnesses.

>> My philosophy is that where possible the transport layer used between
>> my and my neighbor should be invisible to everyone except ourselves.

>A valid opinion.  Not one that I share, however.  Why should such info
>not be a matter of public record, so that time-to-delivery could be properly
>computed?  (Of course, other schemes could be suggested...)

Oops!  I definitely didn't say what I meant there.  How about: "no user
except the two sysadmins concerned should *need* to know the transport
layer I use".  It's not that it should be a secret, it's that it should
not be necessary to know in order to send me mail.

-- 
<David.Keegel@apana.org.au>  <djk@boombox.apana.org.au>  Tel: +61 3 593-1460
aka: werple!tuple!boombox!djk, djk@cs.mu.oz.au.  Soon: djk@bhppmel.bhp.com.au.