On Mon, 24 Mar 2003, Nisan wrote:

> >Can you write up a test-program that seems to repeatedly fail and provide
> >the source code then perhaps I could have a go at it from my end?
>
> I can send you the actual thread that is doing the work. A test program is
> almost impossible, as in the live environment we are doing +-30,000 -
> 100,000 callbacks/day. The urls and host names vary from a few unique hosts
> to 1000's, all depending on what our clients do.

Well, a test program that crashes doesn't necessarily have to be the same
program you're running in your production. It could be just a write-up that
fetches the URL list from a file and feeds your worker-thread or similar.
Whatever that is required for the crashes to occur.

Having a totally separated crashing program also removes the risk that the
bug is induced by something else in your application...

> I can tell you that I dont get the same result when hitting the same very
> (5-10) hosts repeatedly.

That's odd. Given that, can you figure out the lowest number of unique host
names that is required for this bug to appear?

-- 
 Daniel Stenberg -- curl, cURL, Curl, CURL. Groks URLs.
-------------------------------------------------------
This SF.net email is sponsored by:
The Definitive IT and Networking Event. Be There!
NetWorld+Interop Las Vegas 2003 -- Register today!
http://ads.sourceforge.net/cgi-bin/redirect.pl?keyn0001en