Debugging CGI scripts can be a bit of a pain sometimes.

<ul>
<li>you can write a script that filters out the extraneous junk from what's available in Apache's <b>error_log</b> <!-- , that removes almost everything except for just the raw error messages -->
    <ul>
    <li>it's surprising how much extra stuff is added to it, and how readable it becomes after the extra junk is removed
    <li><A href="watch_cgi_errors.pl">this</a> is one such script
    <li>once you start using this script, it's very useful to do this from inside Perl scripts:
        <ul><pre style="margin-top:0; margin-bottom:0; background-color:#afa; padding:0.3em">warn Dumper(\%some_structure) . "\t";</pre></ul>
        This dumps the structure to the error_log only.  The webpage output is unaltered, and there's no indication to the user that any debug activity is occuring.
    </ul>

    <p><br>
<li>the O'Reilly book <A href="http://docstore.mik.ua/orelly/linux/cgi/ch15_03.htm">suggests running your CGI script from the command-line</a>.  That's all well and good, if your script doesn't take any parameters, or is otherwise easy to run from the command line.<br>But if your script is remotely complex, you can use <A href="capture_CGI_input.pm.pl">this script</a> to capture all inputs to a CGI script, when running under a webserver, and allows you to replay that 'input' when running under the command-line later.
</ul>