Steve skrev, on 09-12-2007 23:37:

I could explode!
Don't do that, it's unhealthy ...

For me or the others around me? :)

Basically for yourself, but events in Israel, Irak and Afganistan have proved that it doesn't benefit those around one.

If I enable --enable-debug --enable-bnr-debug and --enable-verbose-debug
then my DSPAM is crashing after some messages are processed. Mostly DSPAM
just segfaults and the socket stays there but the client can not connect to
the socket since DSPAM is not there to respond. But this is not only
limited to the client/server mode. Normal mode does not work as well. Just
crashes.

s/is crashing/crashes/g. English grammar is not German grammar.

Well... in German we do have "is crashing" and "crashes".

In English "is crashing" means it's doing it at this moment (while you watch). "Crashes" means it has a habit of doing so (see http://www.kfmaas.de/presentparticiple.html for a German context explanation).

I have --enable-debug in my standard Fedora/Red Hat rpm spec
with which I (and many others) build dspam 3.8.0. It doesn't
crash (segfault).

I have not tested with simple --enable-debug. Maybe it's the other debug flags 
I used?


[...]

--enable-debug:
  Enable debugging support for DSPAM.

--enable-verbose-debug
  Turns on extremely verbose debugging output. --enable-debug is implied.
--enable-bnr-debug:
  Activates debugging output for Bayesian Noise Reduction.

For --enable-bnr-debug you need to have bnr active in DSPAM.

Ok, so I built an FC6 rpm from stock 3.8.0 source (a couple of non-relevant CGI patches) with --enable-debug, --enable-verbose-debug and --enable-bnr-debug and after a few in-messages dspam still hadn't crashed. I have now got a bnr.log which is useless to me and a load of extra info in dspam.debug which I don't need. So I'll put my original rpm back.

But then on the standard dspam code. Are you building the new, patched stuff, or standard code?

I tried patched stuff. Have not touched standard 3.8.0 code. But I suspect the 
error to be in standard 3.8.0 code.

Well, it doesn't crash on Fedora FC6 and I suspect it wouldn't on RHEL5, which is what I use in production.

[...]

Best,

--Tonni

--
Tony Earnshaw
Email: tonni at hetnet dot nl

Reply via email to