[Top] [All Lists]

Early history and 2821 (was: draft-duan-smtp-receiver-driven-00.txt)

2005-05-13 03:50:59

John C Klensin wrote:

my involvement with ARPANET/ Internet email goes back to the
decision to put a "MAIL" verb in the original FTP

Now that doesn't prove that you're guilty of any attempts to
twist the good old STD 10 reverse path into something like an
arbitrary "bounces to" construct.  Maybe it was somebody else.

For obscure reasons I'm almost always on the wrong list, 2822
issues here, because mail-arch mentions them, and 2821 on the
822 list when authors invent new header fields, last example:

Apparently Bruce found a minor problem in 2821, it's not 110%
clear what to put in the Return-Path for a null MAIL FROM:<>

"former IAB Chair"

Do they discuss issues like "bounces-to" ?  <sarcasm> It's not
very serious, maybe some Terabytes, less than 1 billion USD
damage. </sarcasm>

When it hit me it was in total 150,000 bounces.  With an average
size of 4 KB and my V.90 line that's about 2000 minutes if I got
this right, or 20 EUR for the traffic to me.  Plus the time for
wading through this stuff, sending some complaints, inventing
filters to salvage good bounces and "good" spam (reportable via
SC, last year it didn't allow to report bounces), and solving
several hundreds of these funny C/R puzzles (UOL etc.).

I'm pretty much about as implicated as one can get.  :-)

The crystal balls they sold 16 years ago were really bad, my
exemplar told me that today something like TCP or SMTP would
be ancient history.  All my prophecies were wrong, only SGML
was at least not completely wrong.

Maybe the author of STD 10 had a _working_ crystal ball, his
idea of a "reverse path" was actually correct.  But that was
23 years ago, the quality of crystal balls varies with the
years.  1982 I (ab)used DEV PUN on a VM/370 for local mail,
and I didn't know what a "modem", "ftp", or "smtp" might be.

As soon as time travel is available I intend to discuss
all implications of 5.3.6 (a) and 5.3.3 with the authors.
Maybe the net at large sponsors the trip back to 1989.

Some of those issues where intended to be clarified in 2821.
If you don't think the clarifications were adequate

It's clear from my POV, the only missing piece is to fix it,
back to the working concept of a "reverse path" in STD 10.

this would be a good time to let me know.

That sounds like you're planning a 2821bis anytime soon (?)

In that case I'd like to discuss the 2821bis syntax for an
IPv6-address-literal and a General-address-literal, adding
"IPv6:" is odd, how about adopting the syntax in RfC 3986 ?

It was a pain to find a new syntax for the msg-id in USEFOR,
the "dcontent" in your syntax isn't allowed in news / NNTP,
and it's also "suboptimal" in news: URLs.  After some weeks
we found a working "maximal" syntax.  But your "dcontent" is
defined in RfC 2822, and as always I'm on the wrong list.

        Bye, Frank (only sent to the SMTP list, not ASRG)