httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Thomas Jarosch" <t...@gmx.de>
Subject Re: Help needed: Bug #8480 still present in 1.3.23
Date Mon, 11 Feb 2002 09:57:39 GMT
I tried it with your testfile and it failed.
Then I tried it again on another machine using exactly the
same binaries and there it worked.

I found out that this problem only occurs when I load a special custom module.
The module itself is doing authentication and some content serving.
It installs a "handler" via the "set handler" directive.

The interesting thing is that if I comment out all the references to the
module and just load it via the "LoadModule" directive,
the strange behavior starts to occur.

> d) if actually *nothing* is returned, there *MUST* be a line in the
>    error_log. What does it say? If none is there, try to increase
>    the LogLevel.

It was your hint that there -must- be a line in the error_log.

I've identified the problem: The custom module uses a flag to indicate
if an error has occurred ("errorOccured"). The module checks the flag
in the "[#9] log a transaction" phase of the module and exits
Apache via exit() if it's true.

And now the funny thing: The "errorOccured" boolean
was always initialized with "true". DOH!

Right now I'm wondering why the module ever worked :-)

I apologize for wasting your time and a big THANK YOU!
Tell me which CD you want :-) hehe

cheers,
Thomas



Mime
View raw message