[Spambayes] sb_tray eating time and space with unhandled exceptions

Tony Meyer tameyer at ihug.co.nz
Wed Mar 9 00:43:51 CET 2005


> I have started to see behavior where the sb_tray applet easts up CPU 
> time and generates voluminous log files.

The bug is starting to really annoy me.  Maybe there will be a 1.0.4 after
all (it might be quicker to release that than to answer all the mail about
the problem...)

[...]
> When I look at the large files I find that they have the startup data 
> followed only by *large* numbers of somewhat uninformative 
> lines that read
> 
> warning: unhandled exception
> 
> Do more recent releases provide more information about what these 
> unhandled exceptions are, and can anyone shed any light an 
> what might be happening here?

Unfortunately there is no more recent release.  If you run from CVS Head
(which will be 1.1a1 in the not-too-distant future) then they should only
get written to the log once per hour.  That helps the effect, but doesn't
address the cause, though.

I believe that the unhandled exception message comes from asyncore.  If I'm
following things correctly, select.select() is saying there was a problem
with a socket.  Unfortunately, asyncore doesn't give any more information
than "warning: unhandled exception".  Nor does it look straightforward to
try and collect such information (it looks like a modified/overridden
asyncore.poll() is required).

Since this didn't come up before 1.0.3 (AFAIK) maybe this is something to do
with using Python 2.4, since that's the main change between 1.0.1 and 1.0.3
(and 2.4 did include changes to asyncore).  If you have 2.3 handy and can
run from source, that may fix the problem.  (Or you can use 1.0.1).

I don't really have much time to look into this until next week (and I'd
hoped to use that time to get 1.1a1 out).  I need to either find out how to
replicate this problem (what is it that causes the first error?) or create a
special release that includes a lot more debugging information, and convince
someone to run it.  I might manage to try the latter tomorrow.

=Tony.Meyer

-- 
Please always include the list (spambayes at python.org) in your replies
(reply-all), and please don't send me personal mail about SpamBayes.
http://www.massey.ac.nz/~tameyer/writing/reply_all.html explains this.



More information about the Spambayes mailing list