[ih] The Atlantic on Email

[Dave Crocker]

On 3/15/2016 7:35 AM, Vint Cerf wrote:
> there was also the little problem of inverting the UK domain names from
> uk.ac.ucl.cs to cs.ucl.ac.uk <http://cs.ucl.ac.uk>
> Kirstein's people tore their hair for a while on that and I remember
> some fairly hot debates about this until Internet mail formats (as in
> RFC 722, 822 etc) were agreed as standard.


In those days, this topic always included a debate which I viewed as 
between source-routing, versus canonical, global addressing (which hides 
the routing.)  Standardized- vs. variable-addressing was one of the 
derivative points.  Left-ending vs. right-endian was one of the points 
of variability.

The UK folks used MMDF, which made a point of converting all incoming 
mail to a canonical intrnal addressing model, and then reformulated it 
according to the needs of the outbound channel.  By contrast, I believe 
Sendmail did ad hoc in/out address translation, sustaining the 
combinatorial problems inherent in the approach.

The UK folks had a left-endian addressing scheme, while MMDF internally 
used a right-endian model.  The UK folks modified MMDF to support /both/ 
models.

Expedience vs. purity is always such a painful choice in system design...

d/
-- 

   Dave Crocker
   Brandenburg InternetWorking
   bbiw.net