For a GET, I beleive you can construct multiple CGI.pm objects without this blocking issue.

You can use an alternate constructor for CGI.pm to create more than one without worrying about blocking by passing an empty anon hash:

use strict;
use warnings;
use CGI;
...
# first one, processes POST data if present
my $q = new CGI(); 
...
# passing the anon hash causes this CGI.pm object
# to read its params from the hash instead of
# the environment
my $q2 = new CGI({});
[download]

use strict;
use warnings;
use CGI;
...

my $q = new CGI(); 
...
if( $someRedirectCondition ) {
  my $q2 = new CGI({});
  $q2->param('copy' => $q->param('copy') );
  $q2->param('foo' => 'bar');
  $q2->param('qux' => 'baz');
  ...
  my $uri = "http://some.domain/and/a/path?" . $q2->query_string();
  print $q2->redirect( -uri => $uri );
  exit 0;
}
[download]

my $q=new CGI;
.....
$token=valid_login($q)
redirect_to_login($q) unless $token;
[download]

my $fakeq=new CGI;
$fakeq->{'.cookies'}->{ID1}=$cookieobj1;
$fakeq->{'.cookies'}->{ID2}=$cookieobj2;
.....
$ftokenobj=valid_login($fakeq)
die "Should not happen!" unless $ftokenobj;
[download]

In mod_perl, in some cases you need to create lexically scoped CGI objects rather than a global one (or pass the CGI object into a sub as a reference).

-- Brett

Go not to the Elves for counsel, for they will say both no and yes