[ih] Which two letters? [Internet should be in private hands]

[Brian E Carpenter]

My favourite case was not the notorious cs.ucl.ac.uk, whose mail often ended
up in Czechoslovakia, but uk.oracle.com, Oracle's London office, who supported
CERN when we first became an Oracle customer. JANET, naturally, couldn't
forward to com.oracle.uk, but our gateway thought they could.

We got the heuristics in our mail gateway right in the end, but it took a while.

It was around that time that Jon Postel politely explained that we couldn't
have .cern., and recommended .cern.int., which we declined because politics.

Regards
    Brian Carpenter

On 09-Dec-22 05:06, Olivier MJ Crépin-Leblond via Internet-history wrote:
> Hello all,
> 
> please see my comments interspersed in the text below:
> 
> On 08/12/2022 15:24, Craig Partridge via Internet-history wrote:
>> On Thu, Dec 8, 2022 at 12:59 AM George Ross via Internet-history <
>> internet-history at elists.isoc.org> wrote:
>>
>>> The naming scheme which JANET-attached Universties used did start with
>>> "uk"
>>> rather than "gb" -- we were uk.ac.ed.cs.  That was in place before we
>>> joined the IP internet and started using the DNS too, and I suspect the
>>> prospect of sites having two different names or else having to change the
>>> JANET scheme, which was much more heavily used than the DNS at the time,
>>> didn't appeal, in addition to the political sensitivities.
>>>
>>> We renamed ourselves as dcs.ed.ac.uk shortly after, to avoid confusion in
>>> the mail gateways, as .CS was a ccTLD.
> 
> Yes, a lot of computer departments did so, but the most challenging was
> the other way around, with one of the most influential sites at the time
> cs.net - and I remember that I had the most trouble getting emails to it
> because sometimes it would look-up under .net zone and sometimes under
> .cs zone, so sometimes I used to have to cheat by emailing through a
> third party open relay, like cunyvm.cuny.edu or uunet.net or
> decwrl.dec.com or uucp.sun.com who, for other reasons, all used to be
> pretty good at routing emails cleverly on the Internet.
> 
>>>
>> Quick gloss on George's note for those who weren't there.  The UK, due to
>> JANET or simply a magnetic attraction to doing things the opposite way from
>> most folks, initially had domain names with the TLD first.
>>
>> This meant a bunch of Internet email systems had a "both-ways" switch that
>> involved parsing domain names left to right and right to left to see if
>> either version made sense.
> 
> Only the "gateways" like uk.ac.ucl.cs.nss that then became
> uk.ac.nfsnet-relay, uk.ac.mhs-relay, uk.ac.rl.ib (the RAL gateway to
> Bitnet) and uk.ac.ukc (for UUCP). Trying to do this within the JANET NRS
> scheme (yes that's what's it's called, designed in 1983 - more info on
> https://en.wikipedia.org/wiki/JANET_NRS ) was challenging. I know
> because I tried playing with mailer systems in Coloured Book Software
> and it was impossible, downright impossible to think of every instance
> for it to work well - more info on Colour Book at:
> https://en.wikipedia.org/wiki/Coloured_Book_protocols )
> 
> This is of course not to confuse with the ITU (then CCITT) coloured
> books that determined successive 4 yearly editions of CCITT standards
> like X3, X21, X25, X400 which the UK Coloured Books also used.
> 
> I have no idea why UK was used instead of GB, but the order in which it
> was used probably stems from the X.400 address notation that started
> with Country first, Admin Name, Carrier Name, Organisation Name,
> Recipient Name etc. which in itself, for anyone who has ever used it,
> was nothing short of a nightmare.
> 
> 
>>
>> Eventually, as the DNS grew, some domain names made sense in both
>> directions (stories that some in the DNS leadership team encouraged this
>> mayhem to persuade our friends in the UK to rethink their ordering reflect
>> joking comments at the time, but I don't think anyone did it intentionally)
>> and this led to changes such as those George describes.
>>
> 
> Plus an increasing number of us used the Internet through the NSFNET
> gateway at UCL and got our way finally to convince the powers that be
> that TCP-IP was the future and could be tunnelled over X.25... and then
> of course that was upgraded to native TCP-IP on E1 leased lines soon after.
> 
> Kindest regards,
> 
> Olivier