Skip to Content.
Sympa Menu

devel - Re: [sympa-dev] Loop Prevention

Subject: Developers of Sympa

List archive

Chronological Thread  
  • From: Aumont - Comite Reseaux des Universites <address@concealed>
  • To: Jeff Abbott <address@concealed>
  • Cc: address@concealed
  • Subject: Re: [sympa-dev] Loop Prevention
  • Date: Thu, 16 Feb 2006 09:15:13 +0100

Jeff Abbott wrote:

On Feb 15, 2006, at 2:11 AM, Aumont - Comite Reseaux des Universites wrote:

If needed we can add one more sympa.conf parameter with default value would be 'mailer-daemon|sympa|listserv|mailman|majordomo| smartlist' so you can change it the way you want. It's really easy to do excepted that we have to find a name for this parameter, something like loop_prevention_sender_regexp


Serge,

Thanks for clearing that up for me!

Something occurred to me as I was thinking about this problem -- if this were a configuration variable then we could utilize scenarios to determine whether or not a given list dropped messages from senders matching that variable, couldn't we?

Yes but scenario can't be used to avoid loop in all cases : reply from sympa for unkown commands are not under control of scenario. The incriminated regexp is DoFile subroutine, so it is applied for filtering both list messages for lists and robots messages.

I would propose we pull this hard-coded regex out of sympa.pl, add it as an option to sympa.conf or robot configs,

I agree with this .

set the default scenarios to reject based on that default string, and add an additional action for send.* scenarios: owner, which would send the message in question along to the list's owners (which we would want).

I desagree with changing default scenario. I prefer to give a default value for that new parameter which is eqiuvalent to the previous hardcoded regexp. This way nothing need to be changed on existing installation even if scenario have been customized. Thoses who wants to accept such dangerous mails can use "loop_prevention_sender_regexp .*" and control accepted message customizing some send scenario (may be include.send.header if applied for all lists).

I believe I can produce the code for this using the latest CVS HEAD as my working copy. Hopefully I'll be able to start working on this next week. Does this sound like a good plan and something that will benefit more than just us at Duke?

You are welcome.
Serge



Archive powered by MHonArc 2.6.19+.

Top of Page