|Matus UHLAR - fantomas||May 20, 2010 2:26 am|
|Sam Varshavchik||May 20, 2010 4:04 am|
|Matus UHLAR - fantomas||May 20, 2010 4:52 am|
|Sam Varshavchik||May 20, 2010 3:03 pm|
|Matus UHLAR - fantomas||May 26, 2010 2:38 am|
|Sam Varshavchik||May 26, 2010 4:01 am|
|Gordon Messmer||May 26, 2010 8:32 pm|
|Matus UHLAR - fantomas||May 28, 2010 6:31 am|
|Matus UHLAR - fantomas||May 28, 2010 7:18 am|
|Sam Varshavchik||May 28, 2010 3:08 pm|
|Gordon Messmer||May 28, 2010 4:11 pm|
|Matus UHLAR - fantomas||May 31, 2010 8:24 am|
|Subject:||Re: [courier-users] validating addresses at SMTP time.|
|From:||Matus UHLAR - fantomas (uhl...@fantomas.sk)|
|Date:||May 28, 2010 7:18:20 am|
Matus UHLAR - fantomas writes:
I'm aware of this. Unluckily, some of our customers tend to make many mistakes so the mail bounce (and they quite often misconfigure their source address so the mail double-bounces).
Therefore I would like to push the checks at smtp time.
Currently courier does: - validate sender domain - validate recipient address, if it's local.
What would I like to add: - validate recipient domain - validate sender address
And I'd like to change address validation that it would be validated, if it's known, not only if it's local.
This could give us advantage on MX relays, which could e.g. mirror recipients database using ldap...
On 26.05.10 07:01, Sam Varshavchik wrote:
All of this incurs non-trivial per message overhead. Issuing a DNS query, and/or querying the local account database, is not free, and the penalty now applies to every message. A judgement call needs to be made whether the additional overhead is justified by the value added.
I'm aware of all of this. However all (except sender address verification) the overhead will still happen, but later in delivery chain, in which case it's much harder to tell sender what was the problem with the message, and even if, the overhead in such case is much bigger.
I'm willing and able to accept a little more overhead during the SMTP transaction, because in many cases it will result in much less overhead during later phases of delivery. And you can believe me, having bunch of undeliverable mail in queue causes much more troubles to us than issuing some DNS lookups (1 per recipient, but quite cheap) and database lookups (a bit more expensive, but only one per sender, e.g. mail).
As the old saying goes: build an idiot-proof process, and nature will evolve a better idiot.
Do you mean we shouldn't build idiot-proof processes because of this?
-- Matus UHLAR - fantomas, uhl...@fantomas.sk ; http://www.fantomas.sk/ Warning: I wish NOT to receive e-mail advertising to this address. Varovanie: na tuto adresu chcem NEDOSTAVAT akukolvek reklamnu postu. Chernobyl was an Windows 95 beta test site.
_______________________________________________ courier-users mailing list cour...@lists.sourceforge.net Unsubscribe: https://lists.sourceforge.net/lists/listinfo/courier-users