Bryant, Eric D. <[EMAIL PROTECTED]> writes:
> 5. What MTA do you recommend?
(I think the other questions have been addressed, so I'll just stick to
this one since you'll probably get a lot of "use the MTA that I use"
messages.)
I recommend one of sendmail, exim, or postfix.
qmail has a lot of
Duncan Findlay <[EMAIL PROTECTED]> writes:
> So at a default threshold of 5.0, you could expect an FP rate of
> 0.62%
I think the actual FP rate is much lower than that because we have a
very difficult GA corpus with lots of newsletters and other non-spam
that looks very similar to spam.
It's ve
Porn site spam may include a password in a link that is only good
for a few days. Not that I have ever followed one of them. I just
forward everything like that off to spamcop. Unfortunately, I can't
use spamassassin on my hotmail account.
Dan Liston
Jeremy Kister wrote:
Just autowhitelist th
Actually every user does have a Unix account, but it is separate from
the mail server. They are in the process of changing the whole email
architecture to be a 16-node Linux cluster right now. Should be
interesting to see how it all works out.
-Eric
--On Saturday, October 19, 2002 4:57 PM -04
On Sat, Oct 19, 2002 at 03:01:39PM -0500, Bryant, Eric D. wrote:
> I am working on implementing a spam-filtering solution for Purdue
> University and SpamAssassin is one of the products at the top of my
> list. I'm wondering if you guys can give me some feedback as to what
> your experiences have
--On Saturday, October 19, 2002 4:57 PM -0400 Ross Vandegrift
<[EMAIL PROTECTED]> wrote:
> Watch it so you don't tread on clued users' procmailing. Maybe include
> a warning if the user's .procmailrc already exists, or spit the rules
> out to a different file...
Creative use of environment varia
On Sat, Oct 19, 2002 at 03:01:39PM -0500, Bryant, Eric D. wrote:
> -BEGIN PGP SIGNED MESSAGE-
> Hash: SHA1
>
> Hi,
>
> I am working on implementing a spam-filtering solution for Purdue
> University and SpamAssassin is one of the products at the top of my
> list. I'm wondering if you guys
--On Saturday, October 19, 2002 3:01 PM -0500 "Bryant, Eric D."
<[EMAIL PROTECTED]> wrote:
> 1. Can SA work well as an opt-in/opt-out solution?
I first encountered SA on The Well (http://www.well.com) and it was offered
as an opt-out service. A web page is provided to opt out and to fine-tune
se
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
Hi,
I am working on implementing a spam-filtering solution for Purdue
University and SpamAssassin is one of the products at the top of my
list. I'm wondering if you guys can give me some feedback as to what
your experiences have been thus far with SA
Well, since most people I know aren't stupid enough to type their email in
all caps, I don't have to worry about those getting flagged as spam.
If you've got people who email you in that manner, you might want to
remind them that doing so is akin to yelling, as well as just plain ugly.
On Sat,
> Just autowhitelist the guy. In your ~/.spamassassin/user_prefs (or
> wherever your user_prefs file is located), add this line:
I run SpamAssassin over vpopmail on qmail1.03.. Not only do white lists not
work on an individual popbox, but i wouldnt implement them if they did; It
wouldnt work for
The xxx.xxx is what hit it. That spam phrase searches for text with xxx
xxx in beginning and end (among other things) and it tripped the rule.
On Sat, 19 Oct 2002, Kenneth Porter wrote:
> >From the qpopper mailing list. SA 2.41 thinks it's porn. I can see the
> all-caps "unlimited", but I don't
Bart Schaefer said:
> Use file descriptors. Create a pipe() between the "master" spamd and each
> forked "worker" spamd. The worker closes the read end, the master closes
> the write end, and the worker writes a single byte to the pipe after it
> has processed the message.
>
> The master can
A few days ago, in response to
zenn> how do you guys recommend i setup a system where by our local
zenn> outlook users are about to report spam to razor or to a local
zenn> blacklist
I wrote:
If you're willing to consider a Python-based solution... Mark Hammond
(author of th
>From the qpopper mailing list. SA 2.41 thinks it's porn. I can see the
all-caps "unlimited", but I don't see the porn phrases.
SPAM: Start SpamAssassin results --
SPAM: This mail is probably spam. The original message has been altered
SPAM: so you can reco
I would personally consider the following email spam, but
more to the point, I get very little ALLCAP spams, and the
rules LINES_OF_YELLING and UPPERCASE_25_50 are
more often flagged on legitimate mail. I was wondering if
other people are finding that these rules are actually
effective in flagging
On Sat, 19 Oct 2002, Justin Mason wrote:
> > You were using CVS to do merges?!
> >
> > I wonder if Craig was doing that too. I always create diffs and use
> > an editor, it didn't even occur to me that someone might be trusting
> > CVS to get it right. ;-)
>
> I know, I know -- I'd been spoilt
I was running 2.50 from CVS when I had someone who mentioned to me that
the mail being forwarded from another account at pol.net was being
tagged as SPAM and the AWL score was around 26.3 for them. I reset the
awl db but since I don't know what AWL is looking at, I've just disabled
it altogether.
On Sat, 19 Oct 2002, Bart Schaefer wrote:
> Use file descriptors. Create a pipe() between the "master" spamd and each
> forked "worker" spamd. The worker closes the read end, the master closes
> the write end
Er, sorry, I revised the algorithm in the middle of writing the second
paragraph and
On Sat, 19 Oct 2002, Justin Mason wrote:
> Also, if we can come up with a way to impl a -m-type system without
> counting pids and using REAPER (ie the stuff that dumps core!), then we
> should try that.
No patch, but a suggestion:
Use file descriptors. Create a pipe() between the "master" spam
Daniel Quinlan said:
> You were using CVS to do merges?!
>
> I wonder if Craig was doing that too. I always create diffs and use
> an editor, it didn't even occur to me that someone might be trusting
> CVS to get it right. ;-)
I know, I know -- I'd been spoilt by several years of Clearcase (w
On Sat, Oct 19, 2002 at 12:16:56PM +0100, Justin Mason wrote:
>
> Duncan Findlay said:
>
> > > Where can I get more informaiton on that? I *really* need "-m" to
> > > work. It semed to have been reliably on Linux w/Perl 5.6.1.
> > >
> > > If "-m" is hopelessly broken, does anyone have alternat
* Martin Radford ([EMAIL PROTECTED]) wrote:
> What's interesting about their externally visible servers is that they
> don't do ESMTP; they don't advertise that they do it, and they give
> "500 Unknown or unimplemented command" when you send "EHLO" commands.
> I doubt this is Exim's behaviour (but
At Sat Oct 19 16:40:46 2002, Thomas Hurst wrote:
> Demon use a mixture of exim (externally) and MMDF (internally). It's
> probably exim adding the headers since it gets the messages first.
>
> However, exim 4 generates id's that look like:
>
> <[EMAIL PROTECTED]>
>
> And I don't think exim
* Justin Mason ([EMAIL PROTECTED]) wrote:
> Martin Radford said:
>
> > These are Message-IDs generated by my ISP's incoming mail server for
> > mails that don't already have a message id. And that would explain
> > why no one else is seeing these, while I've got a fair number.
> aha. Yep, we ha
On a related issue, I just upgraded to 2.43 primarily because of the
earlier AWL problems.
My question is, how long will it take for the erroneous AWL entries
to work their way out of my AWL db?
Would I be better off deleting the AWL db and letting it start over?
Thanks,
Ollie
On Fri, Oct 18
On 19 Oct 2002, Lars Hansson moaned:
> On Sat, 2002-10-19 at 01:22, Matt Kettler wrote:
>> if it's 2.43, the AWL tracks both the from address AND the orginating IP.
>
> Uh, I do hope it's the IP that actually delivered the mesage to you that
> is being tracked and not the originating one?
I hope
At Sat Oct 19 14:45:38 2002, Daniel Liston wrote:
>
> The first received: line with (8.9.3/8.9.3) would give me the impression
> that they are using sendmail. [shrug]
No, it's me that's running Sendmail. Demon deliver mail via SMTP to
their dial-up customers (we have static IPs). Only in the la
But if it's the IP address that handed it off wouldn't that be the
system that if forwarding the individuals email to you? That doesn't
make sense to whitelist/blacklist that address because if they are
merely forwarding their mail from that system then it's going to include
spam from that account
Daniel Liston <[EMAIL PROTECTED]> writes:
> The first received: line with (8.9.3/8.9.3) would give me the impression
> that they are using sendmail. [shrug]
No. That is the customer's mail system. Unlike almost all other
dial-up ISPs, Demon offer SMTP mail delivery to dial-up customers.
--
The first received: line with (8.9.3/8.9.3) would give me the impression
that they are using sendmail. [shrug]
Dan
Nix wrote:
On Fri, 18 Oct 2002, Martin Radford stipulated:
I'm pretty sure it's their own custom MTA. The SMTP connection banner
is:
It used to be a distorted MMDF variant, I
On Fri, 18 Oct 2002, Martin Radford stipulated:
> I'm pretty sure it's their own custom MTA. The SMTP connection banner
> is:
It used to be a distorted MMDF variant, I think, but that may have
changed in the last couple of years.
--
`It's hard to properly dramatize, say, the domestic effects o
Justin Mason <[EMAIL PROTECTED]> writes:
> BTW I've stopped using CVS for rule merging, due to this kind of
> brokenness. Easier just to use diff and cut and paste to visually
> verify merging there. So it won't happen again, at least while I'm
> doing it ;)
*GASP*
You were using CVS to do mer
Daniel Quinlan said:
> I already improved this rule back on August 23rd, but someone made a
> broken CVS commit that reverted the improvement (and maybe more, but I
> didn't see anything obvious). I'm starting to wonder if perhaps we
> should have some sort of peer review for back-port and forwa
Duncan Findlay said:
> > Where can I get more informaiton on that? I *really* need "-m" to
> > work. It semed to have been reliably on Linux w/Perl 5.6.1.
> >
> > If "-m" is hopelessly broken, does anyone have alternatives?
> >
> > Note, I'm on sa-talk but not sa-devel. Is that a better plac
Francesco Potorti` <[EMAIL PROTECTED]> writes:
> I intend this to be a guide to people in the same position as me and a
> guide to developers of spamassassin to places in the manual needing
> clarification. It is at
> http://fly.cnuce.cnr.it/spam/spamassassin.html>.
Nice page. Can you file th
"Jeremy Kister" <[EMAIL PROTECTED]> writes:
> here is another email i recieved that was tagged as SPAM... I'm hoping
> someone sees something that can give negative points to this guy ..
> this is my realtor telling me about two houses...
Just autowhitelist the guy. In your ~/.spamassassin/user_
I set up a web page explaining all the things I discovered with
spamassassin while installing it, that is, the things that I needed but
were not on the manual. The page will be updated when I discover
something new.
I intend this to be a guide to people in the same position as me and a
guide to d
38 matches
Mail list logo
