[Mimedefang] pure whitelisting

[Renaud PASCAL]

Le mardi 12 août 2008 16:51, David F. Skoll a écrit :
> jef moskot wrote:
> 
> > Hmmm.  Is there some way to prevent sendmail from bothering MIMEDefang 
in
> > the first place?
> 
> No.  The method I described is the least resource-intensive.

 agreed, still there's another possibility but that'd probably be
tying a shotgun on one's foot to use it in a production environment :-)
 It would require to add-up a sendmail Ruleset that'd take care of
"special" access table entries (e.g. 'WLme:IPaddress<TAB>specialmailer')
and change the mailer for one that'd skip the milter.

 Typically the kind of "add-up" that'll just wait in the dark for
the day you make an upgrade and don't notice the remarks in
the Changelogs about Rulesets and/or mailers and/or milters ;-)


> > If not, is there a simple, default way to just read in a whitelist 
sender
> > and recipient file?
> 
> MIMEDefang offers you a framework... you'd need to implement that
> yourself.  I believe there is code out there that reads access.db
> (which is really just a Berkeley DB file, so you can get at it from
> Perl with "tie".)
> 
> Regards,
> 
> David.

 Correct, there are lots of code about this, db of plain file, easily
found on the web and/or Usenet and much easier to adapt and
maintain than starting to hack sendmail. More, as you point in
another post, a good idea is to conditionnally set flags that will
be analyzed later on to make decision about what to do for which
part of the mail, when dealing with different domains with
different security/commercial rules that's a plus.

-- 
Save a tree: kill a dyzklaimer