[mod_python] Bug in 3.2b5?

To follow up, nothing was wrong with mod_python. Somehow my apache  
conf file had changed and MaxClients was set to 1 in Prefork mode...  
And actually ANY requests to the server was failing... Anyway, sorry  
to have wasted anybody's time on this. I will make sure next time to  
only email the list after a good night of sleep and not past midnight ;)

On a side note, let me take the opportunity to say THANK YOU to all  
the people that make the mod_python community one of the best one  
around! Happy holidays to you all ! Can't wait for 3.2 final ;)

Cheers!

Sébastien


On Dec 20, 2005, at 11:57 PM, Graham Dumpleton wrote:

>
> =?ISO-8859-1?Q?S=E9bastien_Arnaud?= wrote ..
>> Hi,
>>
>> I am running into a pretty bad crash of apache 2.0.55 under OS 10.4.3
>> with mod_python 3.2b5 installed... Before I file a bug though, I
>> would like to know if anybody else has noticed it.
>>
>> I run the following code:
>> [..]
>> 	req.content_type = "application/pdf"
>>     	req.sendfile("/tmp/test.pdf")
>>
>> The file returned in my browser is empty (0K) and I see the following
>> in the apache error_log:
>> [Tue Dec 20 23:20:04 2005] [notice] child pid 983 exit signal
>> Segmentation fault (11)
>
> Are you saying it crashes in "req.sendfile()"? Does it crash on  
> various
> filles or just that one PDF file? How big is that PDF file? Can you  
> send
> the complete handler code, or is that it, bar returning of apache.OK?
>
> It might also help if you go through procedure described in:
>
>   http://www.modpython.org/pipermail/mod_python/2005-December/ 
> 019735.html
>
> to use gdb to find out exactly where it is crashing.
>
>> I tested on my test server (PPC Linux running 3.1.4 and apache 2.0.55
>> as well) and I don't witness the problem, the file is returned
>> properly to the browser.
>>
>> If somebody else is able to reproduce the problem, then I guess we
>> should file a bug.
>
> Lets see if we can work out why first as logging a bug with just this
> information isn't really going to help too much.
>
> Graham