A couple of suggestions:

First, use strict; at the top of the module will keep you out of a lot of trouble.

Next, use CGI; and using it to parse URLs and forms will keep you out of even more trouble. "As simple as possible" is not a good argument. The code you provided would be shorter if you used CGI. (If you're worried about performance, consider CGI::Lite.) In this case, never mind.

Next, if you're running a current (or nearly current) Perl distribution, the date manipulation routines you need should be included, saving you the trouble of lugging around custom methods of uncertain ancestry.

Try something like the following fragment:

  use HTTP::Date qw(time2str);

  my $now = time2str($filetime);
  my $expires = time2str($filetime + $delta);

  print "Last-modified: $now\n";
  print "Expires: $expires\n";
[download]

Finally, one of the subtle problems you'll have is that getting expire times to work depends on a couple of things that might be out of your control.

• The server's clock needs to be set at (or adjustable to) GMT. Expire times are expressed in GMT.
• The browser needs to be set to honor the Expires header. (It's possible to configure most browsers to ignore them, and Cache-Control headers, and refetch every time.)
• The browser's clock needs to be consistent with the server's clock. Good luck here.

Treat setting the Expires header as an optimization, and be prepared to have it ignored.