On 2006-02-28 16:16:33 +0000, [EMAIL PROTECTED] wrote: > I thought there was a message about this in the archive, but I > can't seem to find it...
You probably mean the thread "Should spamhelo DENY_DISCONNECT?" earlier
this month.
> qpsmtpd responds to EHLO with PIPELINING as one of its
> capabilities. Since I am specifically running tests at every stage of the
> SMTP conversation, and there is a strong likelihood that a connection will
> fail before DATA, it doesn't make sense to me to advertise this
> capability.
>
> Perhaps logically, it makes no difference - qpsmtpd will respond
> to the stream with an error at some point, but part of my goal is to
> reduce the bandwidth used.
I haven't actually measured it, but I suspect that PIPELINING saves a
bit of bandwidth. You get one large packet containing MAIL FROM, one or
more RCPT TO (there usually aren't a lot, especially with spam) and DATA
and send back (in the best case) a single packet with the replies.
Without pipelining you receive and send lots of little packets, which
taken together are probably larger.
I would agree that advertising PIPELINING should be optional. You get
somewhat cleaner log files (less "RCPT first" messages) and you could
implement something like earlytalker at each hook.
hp
--
_ | Peter J. Holzer | Ich sehe nun ein, dass Computer wenig
|_|_) | Sysadmin WSR | geeignet sind, um sich was zu merken.
| | | [EMAIL PROTECTED] |
__/ | http://www.hjp.at/ | -- Holger Lembke in dan-am
pgpLeAQJ8n1f9.pgp
Description: PGP signature
