You're somewhat right -- CGI forms can get input from STDIN, but they only get the content that's been sent to them. (you'd normally get this data from a form submission with method='POST') There is also data that's passed in through $ENV{'QUERY_STRING'}, which you'd get if you used a form with method='GET', or if you encoded it manually into a URL, such as:

For older CGI scripts, you'll have to use amperands instead of semicolons to seperate the key/value pairs, but you then have to encode them in the HTML, like so:

To make things even more confusing, you can also pass things in through $ENV{'PATH_INFO'}, by giving a URL with path information past the script, like so

(You don't have to seperate the arguments with slashes -- I don't think there's a set convention for that one)

But, it gets worse when you consider that you could declare a form with

which would result in three different sources of input. And let's not forget that you could also pass extra input in the HTTP headers.

So, what's the solution to all this? Well, use CGI or use CGI::Lite would probably be the easiest, and let them do all the work:

use CGI::Lite;
my $cgi = new CGI::Lite;
my %input = $cgi->parse_form_data();
[download]

Update : missing a 'my'. Doh.