[ILUG] Best Email Server to Use

[Justin Mason]

Colm Buckley writes:
> On Fri, Aug 1, 2008 at 11:30 AM, Michele Neylon :: Blacknight <
> michele at blacknight.ie> wrote:
> 
> > From recollection there was a section about the sender source ie. where the
> > mail came from . This includes other SMTP servers etc., right back to the
> > actual person's desktop.
> 
> And when email is sent through GMail from an SMTP client, all of these
> headers are preserved.  The question remains; what information should be
> captured when the *webmail* interface is used - should we "pretend" that
> it's SMTP and make up a "Received" header for the HTTP client?  That strikes
> me as violating the principle of least surprise, as well as twisting the
> purpose of the header - and I don't like the way Yahoo! does it per your
> example below.

for what it's worth, we in SpamAssassin prefer the Received: header a la
Y! Mail.  those headers preserve ordering nicely, indicating the entire
trail of transmission from start to end (in reverse order).  that's
great for our purposes.

> I still haven't seen a reference to an RFC which we're
> "flouting" with GMail.

I don't think there is one in this case.

> In my opinion, the purpose of the Received: headers is clearly to track the
> mail's progress by SMTP; I really don't think it's good practise to
> arbitrarily redefine it to include webmail receipt.  RFC2821 is a bit
> unclear on the subject, referring only to "mail originating in other
> environments", but I certainly don't see a strong recommendation in any of
> the obvious RFCs to use it for webmail.

This is entirely off-RFC, you won't find anything about webmail.

Michele has a point regarding GMail's current status as a spammer conduit.
It's being massively abused right now, since the CAPTCHA has been broken.
Several very large sites are now blocking GMail; I'll forward details
to you off-list.

--j.