On Fri, 16 Sep 1994 23:03:50 EDT Valdis Kletnieks
<[log in to unmask]> said:
>Why? RFC822, section 4.4.2 says so...
Well, RFC822 doesn't say you have to provide a "Sender:" field if the
message isn't typed from the same workstation that is listed in the
"From:" field. It gives the typical example of secretary typing a message
on behalf of someone else ("when the sender is not the author of the
message", which is not the case here). It then covers the case of the
idiotic support for multiple addresses in the "From:" field, which is
then cleared up by having a single address in the "Sender:" field to say
which of the "From:" authors did really send the message and which are
just co-authors of the message being marked as such in the mail header
for informational purposes, and by the way I'd be interested in knowing
just how many mail programs support that nonsense. Which is, actually,
historically one of the reasons why LISTSERV takes "Sender:" over "From:"
:-)
>What is Listserv's rationale for preferring the Sender: over From: in
>this case?
There are cases, such as a mailing list or a news gateway, where a
message is genuinely coming from user or process X when it was in fact
written by user Y. Doing it this way allows hierarchical private lists,
news gateways, etc. I frankly don't see the point of having a "Sender:"
field if X and Y are the same person. You want a predictable behaviour
for your campus e-mail (usually, you want all users to use the campus
address). If you put two different addresses in the header, some mail
programs/servers will pick one and others will pick another one, and your
helpdesk people will just LOVE it :-)
>As such, MH is merely identifying which entity posted the message, not
>all that far removed from a Received: tag in tracing/auditing.
Except that "Received:" fields are dedicated to tracing/auditing and thus
totally harmless, whereas "Sender:" fields identify a message origin :-)
The root of the problem is that there was no mail server when RFC822 was
written. All this multiple "From:" address business and :; grouping was
meant mostly for human consumption, and as such it is fine. But a mail
server needs a single origin, which had better be something you can
actually reply to. The two main reasons for using "Sender:" over "From:"
are the gateway/list issue I mentioned above, and the fact that "Sender:"
is, in practice, much more likely to contain a working address. It used
to be common for gateways to insert a "Sender:" field with an address
that actually worked, and keep the nice 60-character LAN address in the
"From:" field, with no real guarantee :-) On a more theoretical level, if
the "Sender:" field indicates indeed a secretary sending a message on
behalf of his boss, I see no reason to assume that the boss wants the
automated reply to go to him directly instead of being screened by the
secretary.
Eric
|