Skip to Content.
Sympa Menu

devel - Re: [sympa-dev] Proposal: Milter-Interface for sympa

Subject: Developers of Sympa

List archive

Chronological Thread  
  • From: Patrick von der Hagen <address@concealed>
  • To: address@concealed
  • Cc: address@concealed, address@concealed
  • Subject: Re: [sympa-dev] Proposal: Milter-Interface for sympa
  • Date: Mon, 29 May 2006 18:34:53 +0200

address@concealed schrieb:
[...]
> We agree on the goal to reduce bounces and prefer SMTP code error during
> the session. However we don't think that Sympa have to deal with this.
> Some milter allready exist in order to test if some receipient exists or
> not. j-chkmail, milter-ahead at least provide this service. The only
> configuration where this could prevent a bounce should be if the mail is
> adressed to an existing list hidden to the message sender via a
> visibility scenario. It introduce too much complexity in Sympa for this
> objective. If you provide a contrib we do not garanty that we will
> introduce it into Sympa "official" version because code managment is an
> important issue.
Just to provide some numbers...
Yesterday I received 9525 mails addressed to existing lists at my
Listserv and processed them. An amazing number of 6865 bounces
originated from my listserv, in responso to those 9525 incoming e-mails.
So 2 out of 3 incoming e-mails caused my listserv to send out bounces,
though the recipient-addresses were perfectly valid.

2877 mails have been outright rejected during the SMTP-transaction,
because they were sent to private mailinglists and neither "From:" nor
"Sender:" contained an e-mail-address known to be owner, editor or
member of the mailinglist this message has been sent to. This has been
verified using some kind of "milter".
This milter errs on the side of caution, some more messages could
probably be rejected, but correct interpretation of
Listserv-configuration is just to difficult.

So yesterday the number of bounces sent by my Listserv has been reduced
by 30%, just using this milter. I very, very happy about that. ;-)

Given the trouble spamcop gave me during the last year, I do believe
that a milter interfacing to sympa would be worth the effort. I'll
certainly do a proof of concept.


And of course I understand your concerns regarding code management.
However, DoMessage currently looks (just skimming..) like "setup,
pre_checking, create_context, do_check_list_authz,
acting_upon_list_authz". Bad simplification, I know, sorry.
However, if DoMessage was split like this, an additional Milter-Daemon
might just invoke "setup, pre_checking, create_context,
do_check_list_authz, return_list_authz_to_mta", which might be little
additional overhead for code management. Currently DoMessage is about
250 lines, IMHO that's a rather large method, personally I'd split it in
smaller methods anyhow. ;-)

--
CU,
Patrick.



Archive powered by MHonArc 2.6.19+.

Top of Page