Re: Mandatory From field, anonymity, and hacks

Keith Moore wrote:
> > > The mailboxes (N.B. no named groups) in the From field are used for
> > > replies
> > > (in the absence of a Reply-To field).  So a hypothetical From field with
> > > no mailbox is unworkable in the absence of a Reply-To field, and is
> > > pointless in the presence of a Reply-To field.
> also, From is _not_ pointless in the presence of reply-to because a
> recipient might prefer to reply to the author of the message rather than
> to the address where the author of that message set reply-to.
True, but not quite relevant to what was claimed, viz. that a hypothetical
> From field w/ no usable address is pointless.  If (contrary to RFC 2821)
it is unuseable for replies, then obviously one cannot use it to reply to
the author (Reply-To notwithstanding).

> regarding gateways - in general the job of the gateway is to preserve
> the semantics of the original message into the destination environment.
>  the job of the gateway is NOT to correct errors in the original message.
Well, that's an issue with the RFC 2821 text, which ought to be up for
review Real Soon Now.  It seems that the intent of RFC 2821 is that e.g.
in the case of an X.400 to Internet mail gateway, the address on the
Internet side should be a valid and usable Internet address.  The question
at hand is what should a Usenet news-to-mail gateway do in the presence of
a bogus address (or conversely, should Usenet news, which is an application
of the RFC 2822 Internet Message Format, condone or encourage unusable and
invalid pseudo-addresses in the From field).  As 2821 stands, such a
gateway is required to ensure that addresses are valid and useful for
replies.  As it stands, that effectively precludes Usenet news from condoning
or encouraging invalid and/or unusable addresses. A relaxation of 2821
would affect other types of gateways (e.g. X.400 to Internet mail) as
well.

The current RFC 2822 requirement for a From field provides no opportunity
for anonymous messages, and has led to the sort of hacks that usefor
drafts coondine and encourage.  IT seems to me that relaxing the 2822
requirement provides relief for those desiring anonymity while retaining
the 2821 requirement that addresses in field which are present be valid
and usable (e.g. usable conversion of X.400 addresses).