Re: [irtf-discuss] [Internet Policy] Why the World Must Resist Calls to Undermine the Internet

[Tim Bray <[e-mail]>]

I once spent an unhappy six months very early in my career working in the
IT shop of a fifth-rate telephone company's R&D center, where we had both
Unix and VMS. The VMS C compiler came out and I eventually got UUCP working
on VMS, which I cleverly called UVCP.  I could never make it 100% reliable,
there was something in the VMS terminal driver that randomly lost
characters and I could never get the recovery code to work right. But a
couple tries and your data would eventually go through.  That wasn't the
bad part; the bad part was when our pointy-haired power-crazed IT Director
decided we should go into the software business and started selling it.
Which meant I had to take support calls from the customers who didn't think
that it was OK that it worked except when it didn't. In a rare exhibition
of corporate justice, the IT Director was summarily fired one day.

On Tue, Mar 15, 2022 at 11:05 AM Dave Crocker <[e-mail]> wrote:

> (when searching on 'uucp' be sure to include 'unix', lest you end up
> with pointers to quite a different topic...)
>
>
> As I recall, uucp was indeed universal, quite early in the life of unix.
>
> Routing was manual, source routing.  Hence an 'address' specified each
> node to be transited, producing potentially many exclamation marks in
> the address/route.  And yes indeed, this was at the email level, rather
> than at any lower layers.
>
> Eventually some major switching node emerged, permitting referencing
> them as an anchor, without regard to the path to get to the major
> switch.  As I recall, there were separate apps that computed uucp
> network maps.  But I don't recall how those got used.
>
> FWIW, for CSNet, we used the percent symbol, to indicate one hop, beyond
> the arpanet.  Hence user%csnet-host@csnet-relay.  So, arguably, a hybrid
> of global and relative addressing.
>
> The hybrid -- and the resulting email address syntax -- could get
> complicated -- and error-rone -- when additional email networks were
> part of the transit, such as Bitnet.
>
> Craig Patridge notes significant discussions needed, that produced the
> profound benefit of finally getting agreement among operators of these
> various, independent email services, to use the DNS MX record for routing.
>
>
> d/
>
>
> On 3/15/2022 9:54 AM, Steve Crocker wrote:
> > My nomenclature re uu* is fuzzy.  I know there was a command to copy a
> > file from one Unix machine to another.  I’m not clear on whether this
> > included routing to other machines or whether that was a higher level
> > protocol.
> >
> > Steve
> >
> > Sent from my iPhone
> >
> >> On Mar 15, 2022, at 12:50 PM, vinton cerf <[e-mail]> wrote:
> >>
> >> 
> >> "UUCP routing" versus "UUNET routing"? UUNET did indeed offer UUCP as
> >> its primary service but UUCP was implemented widely on all (?)
> >> UNIX-based systems and likely on non-UNIX systems for interoperability.
>
>