headers
A.M. Kuchling | 16 Jan 2004 02:43
Picon
Gravatar

Compiling elspy on Debian unstable

I'm trying to compile elspy and Exim 4.30 on a machine running Debian
unstable.  The compilation completes without a hitch, but the resulting Exim
binary dumps core when receiving a message.  Even more puzzling, when I run
it under GDB to find the exact location of the segfault, it doesn't dump
core, but delivers the message without problems.

Right now I'm wondering if this is happening because Debian uses a shared
library for Python; perhaps GDB loads the shared library at a different
location and the segfault is therefore avoided.  Has anyone seen anything
like this?  Anyone have suggestions for debugging the problem?

--amk
Permalink | Reply | 4 comments |
headers
Greg Ward | 23 Jan 2004 03:33
Favicon

Re: Compiling elspy on Debian unstable

On 15 January 2004, A.M. Kuchling said:
> I'm trying to compile elspy and Exim 4.30 on a machine running Debian
> unstable.  The compilation completes without a hitch, but the resulting Exim
> binary dumps core when receiving a message.  Even more puzzling, when I run
> it under GDB to find the exact location of the segfault, it doesn't dump
> core, but delivers the message without problems.

Hmmm.  Are you using the latest CVS elspy?  There was an eye-poppingly
stupid bug in elspy 0.1.1 (uninitialized variable) that went unnoticed
for a long time, but then jumped out at me with a coredump when I built
on a new system.  I posted a patch here, but didn't release a new
version because, well, no one seems to be using elspy apart from you,
me, and Neil.  (Sigh.)

(FWIW, the elspy CVS is pretty stable: I haven't touched the code in
months, and both starship.python.net and mail.python.org are running a
CVS version.)

        Greg
--

-- 
Greg Ward <gward <at> python.net>                         http://www.gerg.ca/
I just read that almost half of the population has below median IQ!
Permalink | Reply | 3 comments |
headers
Greg Ward | 23 Jan 2004 03:42
Favicon

test, ignore

Testing list filter -- if this gets through to the list, something is
wrong!

--

-- 
Greg Ward <gward <at> python.net>                         http://www.gerg.ca/
All of science is either physics or stamp collecting.
Permalink | Reply | 3 comments |
headers
Greg Ward | 23 Jan 2004 03:46
Favicon

Re: test, ignore

On 22 January 2004, Greg Ward said:
> Testing list filter -- if this gets through to the list, something is
> wrong!

Damn, something is wrong.

Anyways, I was trying to clamp this list down even more -- I have to
discard 3-5 spams a day for it, and I'm getting tired of that.  I just
want to reject all messages from non-members.

OTOH, elspy is clearly not going to take over the world any time soon,
and I doubt anyone would notice if I just shut this list down.
Objections?

        Greg
--

-- 
Greg Ward <gward <at> python.net>                         http://www.gerg.ca/
If at first you don't succeed, give up--no use making a damn fool of yourself.
Permalink | Reply | 2 comments |
headers
A.M. Kuchling | 23 Jan 2004 03:50
Picon
Gravatar

Re: test, ignore

On Thu, Jan 22, 2004 at 09:46:58PM -0500, Greg Ward wrote:
> OTOH, elspy is clearly not going to take over the world any time soon,
> and I doubt anyone would notice if I just shut this list down.
> Objections?

There might be traffic if my PyCon presentation stirs up some interest --
assuming I can get elspy working.  Why not wait until after PyCon?  If
deleting spam is too irritating, you can turn over the list ownership to me
and I'll handle it.

--amk
Permalink | Reply | 0 comments |
headers
Hy Ginsberg | 23 Jan 2004 14:40
Favicon

Re: Compiling elspy on Debian unstable

Hey, we use elspy here at LiveData!  Elspy rocks!  It's just that, well, it
just simply works, so there's never any need to discuss it...

Hy Ginsberg
LiveData, Inc.
(617)576-6900

----- Original Message ----- 
From: "Greg Ward" <gward <at> python.net>
To: <elspy-users <at> python.net>
Sent: Thursday, January 22, 2004 9:33 PM
Subject: Re: [elspy] Compiling elspy on Debian unstable

> On 15 January 2004, A.M. Kuchling said:
> > I'm trying to compile elspy and Exim 4.30 on a machine running Debian
> > unstable.  The compilation completes without a hitch, but the resulting
Exim
> > binary dumps core when receiving a message.  Even more puzzling, when I
run
> > it under GDB to find the exact location of the segfault, it doesn't dump
> > core, but delivers the message without problems.
>
> Hmmm.  Are you using the latest CVS elspy?  There was an eye-poppingly
> stupid bug in elspy 0.1.1 (uninitialized variable) that went unnoticed
> for a long time, but then jumped out at me with a coredump when I built
> on a new system.  I posted a patch here, but didn't release a new
> version because, well, no one seems to be using elspy apart from you,
> me, and Neil.  (Sigh.)
>
> (FWIW, the elspy CVS is pretty stable: I haven't touched the code in
(Continue reading)

Permalink | Reply | 1 comment |
headers
Gunter Mintzel | 23 Jan 2004 15:18
Picon

Re: test, ignore

Hello everyone, Greg,

 > OTOH, elspy is clearly not going to take over the world any time soon,
 > and I doubt anyone would notice if I just shut this list down.
 > Objections?

This list could be a place to talk about enhancements.

I added some features to the spamassassin.py script, however these are
not documented very well yet. If somebody is interested I would fix this
and release my version somewhere:

- Host and sender whitelisting, to avoid calling SpamAssassin if the
   mail comes from a trusted host AND sender (needs MySQL-DB). This
   saves a lot of CPU time.

- I configured several Anti-Virus rules in SpamAssassin, and don´t
   accept positives even if the REJECT_THRESHOLD is not reached.
   -> less CPU/HDD activity for the AV-Scanner

- The script can post maybe-SPAM mail into several folders of an
   IMAP Server. I use these folders to feed the BAYES db and to
   develop new SA rules.

- The script computes checksums based on whitespace and non-letters,
   to detect SPAM with random word generators (not very useful however,
   needs MySQL-DB).

I use exim, elspy, SpamAssassin 2.6, amavisd-new, Clamscan and a 
commercial AV scanner to process about 20K incoming mails per day
(Continue reading)

Permalink | Reply | 0 comments |
headers
Greg Ward | 24 Jan 2004 03:40
Favicon

Re: Compiling elspy on Debian unstable

On 23 January 2004, Hy Ginsberg said:
> Hey, we use elspy here at LiveData!  Elspy rocks!  It's just that, well, it
> just simply works, so there's never any need to discuss it...

Oh, OK.  I guess very little feedback can be a good thing too.  ;-)
Hope you're not using 0.1.1, because of that nasty coredump bug. I
really need to release a new version one of these months...

        Greg
--

-- 
Greg Ward <gward <at> python.net>                         http://www.gerg.ca/
Question authority!
Permalink | Reply | 0 comments |
headers
A.M. Kuchling | 25 Jan 2004 16:14
Picon
Gravatar

Re: Compiling elspy on Debian unstable

On Thu, Jan 22, 2004 at 09:33:42PM -0500, Greg Ward wrote:
> Hmmm.  Are you using the latest CVS elspy?  There was an eye-poppingly
> stupid bug in elspy 0.1.1 (uninitialized variable) that went unnoticed
> for a long time, but then jumped out at me with a coredump when I built
> on a new system.  I posted a patch here, but didn't release a new

Bingo; thanks, Greg!  Applying the patch fixes the problem.  I have no
explanation for why the code didn't also dump core under GDB; perhaps GDB
fills the stack with zeros or something?  Getting a 0.1.2 release out before
PyCon would probably be a good idea.

--amk
Permalink | Reply | 0 comments |

Gmane