Matt Sergeant wrote:
MS> Well personally I'd rather see the test harness code improved to time regexps
MS> too, and alert of slow ones. However that requires a lot more work on the
MS> behalf of the developers who have so far completely ignored the test
MS> framework I wrote.
I've implemented so
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
On Thursday 25 Apr 2002 6:34 pm, Craig R Hughes wrote:
> Matt Sergeant wrote:
>
> MS> Now that's interesting. I wonder if it's constructed specifically to
> make SA MS> hang... I guess we need to redo the body test parser to grep
> out blank lines. MS
On Thu, Apr 25, 2002 at 02:12:27PM -0600, Charlie Watts wrote:
> > Doing an alarm() and such would side-step the problem, but not actually
> > stop the problem from occuring.
>
> How about both: regexes -are- complicated, and folks are likely to
> write long-running ones again in the future.
>
>
On Thu, Apr 25, 2002 at 02:12:27PM -0600, Charlie Watts wrote:
> On Thu, 25 Apr 2002, Theo Van Dinter wrote:
>
> > On Thu, Apr 25, 2002 at 11:31:08AM -0600, Charlie Watts wrote:
> > > Is there anything that can be done to -avoid- this?
> > >
> > > Perhaps wrapping the regex portion of SA in an al
On Thu, 25 Apr 2002, Theo Van Dinter wrote:
> On Thu, Apr 25, 2002 at 11:31:08AM -0600, Charlie Watts wrote:
> > Is there anything that can be done to -avoid- this?
> >
> > Perhaps wrapping the regex portion of SA in an alarm() timer so that
> > long-running regexes just get exited from?
>
> I vo
On Thu, Apr 25, 2002 at 11:31:08AM -0600, Charlie Watts wrote:
> Is there anything that can be done to -avoid- this?
>
> Perhaps wrapping the regex portion of SA in an alarm() timer so that
> long-running regexes just get exited from?
I vote for "writing correct code the first time", followed by
Matt Sergeant wrote:
MS> Now that's interesting. I wonder if it's constructed specifically to make SA
MS> hang... I guess we need to redo the body test parser to grep out blank lines.
MS> *sigh*
Probably be interesting to count the ratio of blank:nonblank lines and give it a
score.
C
On Thu, 25 Apr 2002, Matt Sergeant wrote:
> On Wednesday 24 Apr 2002 8:02 pm, Viraj Alankar wrote:
> > Thanks, I've made these changes and the memory usage does not seem bad. It
> > still grows a little, reporting about 26M via top, but definitely not as
> > bad as before. I will do some tests wit
On Thu, Apr 25, 2002 at 11:12:17AM +0100, Matt Sergeant wrote:
> > http://www.cse.fau.edu/~valankar/blah.gz
> >
> > Didn't debug this much, but I did disable network checks and still the same
> > hang result with this message.
>
> Now that's interesting. I wonder if it's constructed specifically
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
On Wednesday 24 Apr 2002 8:02 pm, Viraj Alankar wrote:
> Thanks, I've made these changes and the memory usage does not seem bad. It
> still grows a little, reporting about 26M via top, but definitely not as
> bad as before. I will do some tests with c
On Wed, Apr 24, 2002 at 04:53:50PM +0100, Matt Sergeant wrote:
> On Wednesday 24 Apr 2002 4:12 pm, Viraj Alankar wrote:
> > $sa_obj is re-used for every new message that is scanned. Running on
> > Solaris 450mhz sparc system and perl5 (5.0 patchlevel 5 subversion 3), I've
> > noticed from 'top' th
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
On Wednesday 24 Apr 2002 4:12 pm, Viraj Alankar wrote:
> $sa_obj is re-used for every new message that is scanned. Running on
> Solaris 450mhz sparc system and perl5 (5.0 patchlevel 5 subversion 3), I've
> noticed from 'top' that the memory usage grow
Hello,
I am using the Mail::SpamAssassin v2.20 in a content filter for Stalker
Communigate Pro mail server and have some questions.
My filter is basically working like this:
my $status;
my $body;
...
$status = $sa_obj->check_message_text($body);
$retval = $status->is_spam();
$status->finish();
13 matches
Mail list logo
