# Output an image to the browser
sub show_img {
 my $img = shift;
 my ($mime) = $img =~ /\.(gif|jpe?g|png)$/;
 $mime = 'jpeg' if $mime eq 'jpg';
 open my $fh, '<', $img or die "Could not open $img for read: $!";
 binmode $fh;
 my $img_data = do { local $/; <$fh> };
 close $fh;
 print header('image/' . $mime), $img_data;
 exit;
}
[download]

print redirect("/some/webpath/to/selected/image.jpg");
[download]

• The server is better than you at pumping data.
• The values of last-modified and if-modified-since will be respected, allowing the server to avoid sending the data if the browser already has the image cached.
• You won't ever be able to accidentally access something outside the normal htdoc tree, even from bad programming or malicious hacking.
• Your CGI process goes away quicker.
• The code will be eight lines shorter. {grin}

-- Randal L. Schwartz, Perl hacker
Be sure to read my standard disclaimer if this is a reply.