For errors that occur after you've already spit out the header, you can write a custom handling routine (e.g. use or bail_gracefully($!) instead of die, and have bail_gracefully do whatever needs doin' with $!.)

Most HTTP servers have a facility for generating custom error pages -- you can even set up a CGI to handle them. So read up on your webserver's documentation. Under Apache, it can be as simple as adding the line

ErrorDocument 500 /cgi-bin/error_handler.pl
[download]

Which says "if there is an error with code 500 (caused by a program's failing to execute, or to output proper headers), redirect the user to this URI."

HTH

perl -e 'print "How sweet does a rose smell? "; chomp ($n = <STDIN>); 
+$rose = "smells sweet to degree $n"; *other_name = *rose; print "$oth
+er_name\n"'
[download]

One note on the ErrorDocument 500 method: make sure if you use a .pl script for this rather than a static .html file that the script is guaranteed to execute flawlessly, or you get into a potentially nasty situation.

Beware signals..
It's quite possible that the signal being handled at a 'bad time' will actually cause the process to segfault, which is worse than the initial behaviour that's being circumvented.

Malk

Create an error routine to execute when something goes wrong, present a happy screen to the users letting them know that some sort of non problematic (don't panic!) problem has occured and to contact X if they desire.

then internally, log what really happened (the web logs might suffice instead), or perhaps shoot an email to yourself with the real error so you can dig in and fix it.

for instance, if we can't connect to the database, rather than present the ORA error messages, we just post a "system down" message that assures the users that it's temporary, and provides information for them to contact support.

good luck!

Hi, this is the code I use in my scripts. I use DieNice instead of die. You can save recoding all your dies by setting a $SIG{__DIE__} = \&DieNice as noted above. In my experience the die sig handler is not reliable on Win32.

sub DieNice {

    my $message = shift;
    my ($package, $file, $line) = caller();
    my $user_message = "Sorry, the system is currently unable to proce
+ss your request<br>\n";
    $user_message .= "due to routine maintenance. Please try again lat
+er. Our Apologies\n";
    $message = Unindent <<"    MESSAGE";
    A fatal die was trapped by the $scriptname die nice routine, detai
+ls:
    Time:          $datetime
    List name:     $list_name
    Script name:   $scriptname
    Package:       $package
    File:          $file
    Line Number:   $line
    Error Message: $message
    MESSAGE
    &TellUser($user_message);   # the bullshit message!
    &WarnAdmin($message);       # the real facts!
  exit;
}

sub WarnAdmin {

    my $message = shift;
    my ($old_name, $old_email) = ($name, $email);
    $name = "Administrator";
    $email = $our_email;
    &EmailUser ($message);
    ($name, $email) = ($old_name, $old_email);
  return;
}
[download]

The &TellUser sub displays a Sorry page with the passed message. The old routine maintenance story!. The unindent sub just strips leading whitespace and lets the herepages sit nicely inline in the code.

cheers

tachyon

s&&rsenoyhcatreve&&&s&n.+t&"$'$`$\"$\&"&ee&&y&srve&&d&&print

open (FILE, "/usr/tmp/doesnt/exist") 
   or print "Location: http://www.perlmonks.org\n\n";
[download]