also: when you get the "premature end of headers" error, the webserver's error log can contain additional useful information (like what is received when it expects the header - generally the error is triggered by code that sends out messages/content before/while sending the header block)

update: you should never use &subname; without parentheses to call a subroutine unless you know why that's a special case, and if you're using parentheses it's always safe to leave off the &, so IMO it's generally a good habit leave off the & and use parentheses where it makes sense, like subname()

The error log for my server just gives
[Wed Sep 19 13:42:19 2007] [error] [client xxx.xxx.122.100] Premature end of script headers: consent.cgi, referer: http://mysite.com/ase-workkeys/employerconsent.html
over and over again.

As for the subroutine calling, I used to call them with the () and no &, but someone told me that was unnecessary if the subroutine did not return any values. What is the danger? I am asking seriously. Thanks.

---

I learn more and more about less and less until eventually I know everything about nothing.

The danger is that calling a sub as &subname; (with an ampersand, without parentheses) will pass on the current subroutine's @_ argument array. Which is almost never what you want to do.

Using both parentheses and & sign is the same as just parentheses, and you can usually get away without parentheses too, provided the subroutine is defined before the call.

"What should it profit a man, if he should win a flame war, yet lose his cool?"

did you mistakingly change ownership to root? for example, deleted and recreated consent.cgi? Also, is consent.cgi chmod 0755 ? do ls -lha /your/path/to/consent.cgi

oh wow.. ok .. you've got it easy on one hand because you have full access.. and hard on another because you have full access.

do this instead.. vim /usr/bin/debughttp (as root)

#!/bin/sh
# this is just an example
#tail -f /var/log/httpd/error_log
tail -f /var/log/httpd/ssl_error_log | sed 's/^.\+0.41\]\|\, referer.\
++\.cgi\|\?rm=\w\+//g'

# or just 
# tail -f /var/log/httpd/error_log
[download]

Open a terminal and # debughttp forget the CGI::Carp 'fatalsToBrowser' crap.. get rid of all that, get rid of the open(STDERR, '>>' .. all that.

Make sure you have the right path to your httpd error log

call the script from your browser and enjoy the wonder that is POSIX.

Ok, wow. Yeah, I may have root, but I don't understand any of that, so I'm not going to run off and put that on my server until I can spend sometime trying to figure out what tail does.
In Apache that user is set up to have the error log at /home/account/logs, is this not the httpd error log?
Thanks.

UPDATE: In the /var/log/httpd/suexec.log, I found this:


[2007-09-19 14:00:05]: uid: (506/devpreview) gid: (506/506) cmd: conse
+nt.cgi
[2007-09-19 14:00:05]: command not in docroot (/home/acccount/cgi-bin/
+asewk/consent.cgi)
[download]


I assume this is a user account issue then and not my actual perl script?

---

I learn more and more about less and less until eventually I know everything about nothing.

If your answer to what I posted is '..trying to figure out what tail does..' Maybe you should not have root access on a server that serves outsite an intranet.

You should know to # man tail

You should really learn unix before you have that kind of control over a server.

If you know what you're doing, you can secure a machine like that to be a fortress. If you don't- and you have root.. you can blow yourself up- maybe even take some people with you.

I don't think there's any way I can prevent as coming accross as patronizing. I am sorry for that. I do not mean disrespect.

I am sure you can learn this up and down and left and right. But until you do- I must stress - don't open that thing to accept outside requests- via the internet.

A week or two of reading up could do a lot for you.