Lars Magne Ingebrigtsen | 1 Feb 2010 11:34
Face
Picon
Favicon
Gravatar

Re: Further developments

Duncan <1i5t5.duncan <at> cox.net> writes:

> Huge files?  I thought you were doing a file per message? 

I was just testing with dd if=/dev/zero...

--

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi <at> gnus.org * Lars Magne Ingebrigtsen
Lars Magne Ingebrigtsen | 3 Feb 2010 16:21
Face
Picon
Favicon
Gravatar

NNTP spam filtering server side

I just had an idea.  How about if the news server were to leave articles
cross-posted to gmane.spam.detected out of the overview data?  Then
people who read via the nntp interface would get a much less spammy
experience than they have today.

I could just hack up the function in inn that outputs the overview data
to check for that group in the lines it outputs.

In addition, I could have the server listen to two addresses.
If you read from complete-news.gmane.org instead of news.gmane.org,
you'd get complete overview files, instead of the censored ones.

What do you think?  Would this be worthwhile?  

--

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi <at> gnus.org * Lars Magne Ingebrigtsen
Duncan | 4 Feb 2010 02:28
Picon

Re: NNTP spam filtering server side

Lars Magne Ingebrigtsen posted on Wed, 03 Feb 2010 16:21:49 +0100 as
excerpted:

> I just had an idea.  How about if the news server were to leave articles
> cross-posted to gmane.spam.detected out of the overview data?

I like! =:^)

--

-- 
Duncan - List replies preferred.   No HTML msgs.
"Every nonfree program has a lord, a master --
and if you use the program, he is your master."  Richard Stallman
Olly Betts | 4 Feb 2010 08:37
Favicon
Gravatar

Re: NNTP spam filtering server side

On 2010-02-03, Lars Magne Ingebrigtsen <larsi <at> gnus.org> wrote:
> In addition, I could have the server listen to two addresses.
> If you read from complete-news.gmane.org instead of news.gmane.org,
> you'd get complete overview files, instead of the censored ones.

I'd certainly prefer to be able to read the "complete" version myself - I don't
see that many spam messages which gmane has cross-posted, and do see the odd
mis-marked message which I like to be able to report as "unspam" (perhaps one a
week - most recently was today).

But I can see it being a nicer default.

Cheers,
    Olly
Joe(theWordy)Philbrook | 4 Feb 2010 14:03

Re: NNTP spam filtering server side


It would appear that on Feb 4, Olly Betts did say:

> On 2010-02-03, Lars Magne Ingebrigtsen <larsi <at> gnus.org> wrote:
> > In addition, I could have the server listen to two addresses.
> > If you read from complete-news.gmane.org instead of news.gmane.org,
> > you'd get complete overview files, instead of the censored ones.
> 
> I'd certainly prefer to be able to read the "complete" version myself - I don't
> see that many spam messages which gmane has cross-posted, and do see the odd
> mis-marked message which I like to be able to report as "unspam" (perhaps one a
> week - most recently was today).
> 
> But I can see it being a nicer default.

Like you Olly, I don't see so much actual spam on any of the lists I read
as to bother me all that much. But what I like is the idea that the users
will be able choose which feed. And this dual nntp address method makes
that so easy as to not much matter which is the default.

I'm curious though, because I normally use nntp rather than http, I'm not
to sure of the web access options. If I chose to keep my news reader
pointed at the spam filtered news.gmane.org address, what would I need to
do to insure when I occasionally use http://search.gmane.org/ to find
something, that I would be searching the "complete" version???

Also, I never did look into how the spam filter works. Lets say that
Spambot(at)sicko.ru succeeds in posting to a mailing list, and that gmane's
filter excludes it from the news.gmane.org mirror of that list. Would any
list replies to the spam also be filtered out??? (you know the ones from
(Continue reading)

Olly Betts | 4 Feb 2010 21:49
Favicon
Gravatar

Re: NNTP spam filtering server side

On 2010-02-04, Joe(theWordy)Philbrook <jtwdyp <at> ttlc.net> wrote:
> I'm curious though, because I normally use nntp rather than http, I'm not
> to sure of the web access options. If I chose to keep my news reader
> pointed at the spam filtered news.gmane.org address, what would I need to
> do to insure when I occasionally use http://search.gmane.org/ to find
> something, that I would be searching the "complete" version???

The web UI has been already hides anything cross-posted to gmane.spam.detected.
And because of that, the search doesn't index such messages (there wouldn't
be a URL to link to in the results).

> Also, I never did look into how the spam filter works. Lets say that
> Spambot(at)sicko.ru succeeds in posting to a mailing list, and that gmane's
> filter excludes it from the news.gmane.org mirror of that list. Would any
> list replies to the spam also be filtered out??? (you know the ones from
> those list members who seem to think there's a point to publicly telling
> the sicko who posted the spam what they would like to do to him/her if
> he/she doesn't stop spamming the list...)

Not automatically, though if a reply quotes the whole of the spam message it's
likely spam assassin's Bayesian classifier will flag the reply as spam if they
thought the original was.

Cheers,
    Olly
Joe(theWordy)Philbrook | 5 Feb 2010 04:41

Re: NNTP spam filtering server side


It would appear that on Feb 4, Olly Betts did say:

> The web UI has been already hides anything cross-posted to gmane.spam.detected.
> And because of that, the search doesn't index such messages (there wouldn't
> be a URL to link to in the results).

So then, I suppose that on those rare occasions when I'd like to include
those messages in my search, I could run the search criteria twice, only on
the 2nd search specify "gmane.spam.detected" in the "In group" field...
(or is it manifestly impossible to "search" gmane.spam.detected???) I wonder
if it would be asking too much to suggest an additional search criteria switch
to include matches that are "cross-posted to gmane.spam.detected"?
(With the default, of course, being to exclude them...)

 
> > Lets say that Spambot(at)sicko.ru succeeds in posting to a mailing list, and
> > that gmane's filter excludes it from the news.gmane.org mirror of that list.
> > Would any list replies to the spam also be filtered out???

> Not automatically, though if a reply quotes the whole of the spam message it's
> likely spam assassin's Bayesian classifier will flag the reply as spam if they
> thought the original was. 

<sigh> Seems to me that debates and laments about the spam are at least as
annoying as the spam itself... Still I suppose it would take a bit of work
to automatically flag the entire subthread based on any replies to spam inserted
into a legitimate thread without stepping on too many legitimate replies to
the original thread. So I return to my original opinion that I don't
personally see enough spam to worry about it all that much. </sigh>
(Continue reading)

Daniel Shahaf | 14 Feb 2010 20:54

Re: g.c.v-c.subversion.announce getting subclipse-announce posts?

Lars Magne Ingebrigtsen wrote on Fri, 29 Jan 2010 at 18:38 +0100:
> Daniel Shahaf <d.s <at> daniel.shahaf.name> writes:
> > Does something need fixing?
> 
> Yes -- apparently three different lists were funnelled into the same
> group.  I'm now unsubscribing the other two and resubscribing under new
> addresses. 
> 

The problem still occurs:

http://permalink.gmane.org/gmane.comp.version-control.subversion.announce/300
Arioch | 16 Feb 2010 20:46
Picon
Favicon

Gmane as tool for spam address harvesters ???

Hello!

Look at http://s005.radikal.ru/i211/1002/55/39d129277e58.png

Gmane requires us to give real e-mail addresses, and - and passes them  
un-mangled  in plain text :-/

--

-- 
Написано в почтовом клиенте браузера Opera: http://www.opera.com/mail/
Adam Sjøgren | 16 Feb 2010 21:56
X-Face
Picon
Favicon
Gravatar

Re: Gmane as tool for spam address harvesters ???

On Tue, 16 Feb 2010 22:46:43 +0300, Arioch wrote:

> Gmane requires us to give real e-mail addresses, and - and passes them
> un-mangled  in plain text :-/

http://gmane.org/faq.php

  Best regards,

--

-- 
 "Lef ma nine imma Jeep"                                      Adam Sjøgren
                                                         asjo <at> koldfront.dk

Gmane