On Wed, 29 Sep 2004, Matthew Elvey wrote:
Pretty much what Danny Angus said:
I strongly prefer a standard which adds mandates that improve
consistency, and mandates existing optional behaviour, rather than
duplicates the optional with a mandatory alternative.
Instead of MAY add some piece of data to an existing header type, make
it SHOULD or MUST instead and already those people who implement the
option will be compliant with that new thing.
Did I ever say this was a mandatory or alternative to anything?
Just like information put in the received lines is not mandatory but
is considered to be "good behavior" so would this be good behavior
for systems that change source and destination addresses in transit.
It is definetly not going to be MUST at best SHOULD for old values
and MAY for new values (although i prefer SHOULD...)
I also note that information provided by the construct is currently
not available at all. This is its value, as it allows to see changes
made to email source and destination in consistant way.
As far as amount of data, over 50% of emails (especially the shorter ones)
are user to user (source to destination) direct and would not have this
line, the other emails mostly pass through one or two forwarding hops
which already add tons of Received and X- lines, at best this would add
25% extra header data to such emails. And in any case, all this data is
completely ignored by most end-users who are not techs and are not
interested in how email got to them.
Also all email signing systems will add A LOT MORE data to emails just
by the fact that full signatures takes a lot more space. I'm not against
it and infact based on the amount of email data that ISP processe, it
seems that percent of SMTP traffic over overall traffic decreased (inspite
of increasing amount of spam). What happened is that ISP infrastructure
has been upgraded with a lot bigger pipes both on the ISP networks and to
end-users who now often have DSL and cable and email use which was big
deal during dialup still remains but for DSL subscriber the real use
of internet is now P2P, MP3 downloads, gaming, HTML surfing with a lot
more graphics, etc. So we should not see ideas that cause increase in
size of emails as big deal for network infrastructure - it'll have
minimum impact on it and most it could is increase size of email box
storage but with advancement of storage devices and hard drives of
300GB now on the market (as opposed to 30GB 5 years ago), this also
woule have no serious impact.
--
William Leibzon
Elan Networks
william(_at_)elan(_dot_)net