Hello,

I announced today a project I had in mind since last July, which aims at 
helping XMPP server operators fight spam in a *proper* way.

The project is called Providence, and aims at being a generic daemon that can 
be plugged to any XMPP server via a server module. Works similarly to 
SpamAssassin for email.

Project on GitHub: https://github.com/valeriansaliou/providence 
<https://github.com/valeriansaliou/providence>
Post describing & explaining the project: 
https://journal.valeriansaliou.name/announcing-providence-a-spam-filter-for-xmpp/
 
<https://journal.valeriansaliou.name/announcing-providence-a-spam-filter-for-xmpp/>

Providence will implement a set of filters that are known to work well for 
email, and I suspect will also work decently well for XMPP (see Features 
category on GitHub in the README).

As I am very busy at the moment on other projects, I postponed the release of a 
first working testing version of Providence to the end of the first semester of 
2017.

If you have any idea to suggest / or you want to discuss about an idea I 
proposed in the project/article, I’d be more than happy to discuss this there!

--

Valerian Saliou

> On Nov 19, 2016, at 10:40 PM, Dave Cridland <[email protected]> wrote:
> 
> On 19 November 2016 at 13:27, Krzysztof Grochocki <[email protected]> 
> wrote:
>> Hello.
>> 
>> Over the past year I received spam message only in russian language or in
>> russian and english language together. I think we can block such messages
>> like it is in one of polish IM - just block incoming/outgoing messages where
>> is Cyrillic characters in text. At this time it's only the one way to stop
>> spam. Maybe someone can write such functionality as module for ejabberd and
>> prosody?
> 
> It works; it's even useful for some deployments.
> 
> https://github.com/dwd/Metre/blob/master/src/filters/unicode.cc
> 
> Setting that just to block cyrillic blocks is pretty good for me, actually.
> 
> But it's obviously not a universal solution.
> 
> SPIM is hard because they're very short messages and the intention is
> to forward along to the client (more or less) instantly. So unlike
> email, where we can delay to see if we get other, similar messages -
> and note that even this is only useful for a large server - or delay
> while we do some enhanced lookups (RBL checks on URLs, for instance)
> delay is going to block a considerable amount of traffic due to the
> in-order rule, which in turn introduces a potential DoS.
> 
> For personal servers - and we really want to encourage these - there's
> very little we can do, and the risk of deploying a quick fix like the
> above is that wide deployment would simply progress an arms race.
> 
> Dave.

Attachment: smime.p7s
Description: S/MIME cryptographic signature

Reply via email to