Am Thu, 24 Mar 2011 13:11:24 +0100
schrieb Ichthyostega <[email protected]>:

> 
> Hi all,
> 
> recently, some new spammers seem to have discovered our Trac :-P
> As a first "escalation step", I've now switched on eMail verification,
> just to make their life a little bit harder.
> 
> Basically, I'm rather not much concerned, because our project is small
> and we don't yet have the need for users to report bugs.
> (unfortunately!)
> 
> Next things I'd propose to try would be to upgrade our Trac to 0.12
> and then install the spamfilter trac plugin (which I know to work
> reasonably well, unless the installation is really put under high
> pressure by the spammers)
> 
> Any other thoughts, suggestions?

some time ago i created a spamfilter daemon (rxpd) which was planned to
be used for different things (uwiki for example). I also use it as
enhanced IRC-Troll filter :).

It has a very simple textual protocol for a client to get anything
checked against various lists of regular expressions (it is possible
to maintain different lists for different purposes).

Writing a plugin/client for trac should be quite simple, basically one
just needs to send the data to be checked to the rxpd and parse its
response (which is either the regex matched or a '#OK:' string).

I suggest rxpd, because it works and because it has some more benefit
as we can use it for more than just the trac, filtering Mailinglist
posts (there is increasing spam too), uwiki someday and so on.

Moreover the cinelerra trac gets spamed too and sooner or later they
maybe want to move the cinelerra ML to the server. One single rxpd
instance can serve them all.

Rxpd was based on the experience i made with the MoinMoin wiki long
ago. It had one single list of regex, sycned from a master server, which
initially worked quite well but it had some shortcomings. First the
list grown larger and larger and since it was dynamic every cgi call
had to recompile the regexes which became rather slow, second there
was no kind of profiling/timestamping you have no idea which rules
trigger frequently and which are useless cruft. Rxpd addresses this by
being a daemon on its own which compiles all regexes only once and can
serve requests of arbitary clients in a very fast way (I planned to
change the regex engine from the libc/posix regex to libtre which is
awesomely fast, but no need yet) and it maintains 'atimes' for each
rule and has the ability to expire unused rules. Further it can also
work in a Peer2Peer mode where it can merge Filter lists from
other sources, unfortunally this never got flying because it seems i am
the only user of my Software :P guess it needs some more
advertisements. 


The README:
 
http://git.pipapo.org/?p=rxpd;a=blob;f=README;h=fa9e06bf6a3f45714861093ece630a72f0721284;hb=db22714b86113b92fd9595b269c5997a59e0fdca

(the demonstration mentioned at the end is currently not running)


Well the catch is that I am not a python guy and don't want to dig in
Trac and Mailman sources to hook this up. Someone else should do this
(I can assist him with any rxpd related things and also setup an rxpd
on the server)



        Christian


> 
> Cheers,
> Hermann V.
> 
> 
> 
> 
> _______________________________________________
> Lumiera mailing list
> [email protected]
> http://lists.lumiera.org/cgi-bin/mailman/listinfo/lumiera
> http://lumiera.org/donations.html

_______________________________________________
Cinelerra mailing list
[email protected]
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Reply via email to