Re: [yam] Issue #10: RFC 5321 3.9: add tiny subsection

[Alessandro Vesely <[e-mail]>]

Hi SM,
you wrote:
>  From Section 2.1:
> 
>   "In other words, message transfer can occur in a single connection
>    between the original SMTP-sender and the final SMTP-recipient, or can
>    occur in a series of hops through intermediary systems."
>
> I gather that you are viewing "forwarding" as when the final 
> SMTP-recipient being changed by an intermediary system.

These cases are interesting for privacy issues. Since it is legally 
different if I send to a list rather than type or select from my 
personal address book each recipient, one may warn that by changing 
envelope addresses an intermediary server may violate the final 
recipients' privacy. Would that belong to the Security Considerations 
section, then?

> In technical 
> terms, we could have different types of forwarding; for example the 
> address correction in Section 3.4.  Section 3.7.2 also discusses about 
> forwarding where the message goes through a gateway.  From an 
> implementation perspective, we might talk about aliases, as mentioned in 
> Section 3.9.1, or the .forward which can be set by a user.  There's also 
> the case of this mailing list which forwarded your message to me.  
> Section 3.9.2 refers to that as "redistribution".

In plain English, any hop should be considered a case of SMTP 
forwarding, with the possible exception of gateways which, by their 
nature, talk SMTP on one side only. Out of SMTP scope, "forwarding" 
also means users manually resubmitting messages.

> The main problem doesn't seem to be with forwarding; it's more about the 
> return path and/or authorization.  It's up to the WG to determine 
> whether restating all that would be a minimal change.

You mean SPF authorization?

Return-path discussions are often cluttered with source routes issues. 
IMHO the full standard should consider source routes an historical 
oddment that nobody cares about, any more. Cleaning that up might not 
be a minimal change, but would improve readability.