I like this idea, but it has the major drawback that a hash dosen't preserve order, and in this case the order of application of those REs is important.

I don't know of any elegant solution to this (I'm using mostly the same method when creating multi-page CGI scripts), so the thing I'm using is a hash of page handlers and an array as the sequence of pages (yes, it's the dreaded wizard-style interface). Pages (== hash keys) that are not found in the sequence array are then either flagged as an error (more stability) or silently appended to the end of the wizard (quick'n'dirty development).

my @pageorder = ( "welcome", "address", "recipients", "preview", "done
+" );

# Routines to create the pages
my %pages = (
  "welcome" => \&page_welcome,
  "address" => \&page_address,
  "recipients" => \&page_recipients,
  "preview" => \&page_preview,
  "done" => \&page_done,
);

# Routines to validate user input for every page
my %validators = (
  "welcome" => undef,
  "address" => \&validate_address,
  "recipients" => \&validate_recipients,
  "preview" => undef,
  "done" => undef,
);
[download]

And the page handler works like this :

# First, determine where we are, using a default of "welcome" on error
+ :
my $page;
if ($cgi->param("go_prev") and $cgi->param("prevpage")) {
  $page = $cgi->param("prevpage");
} else {
  $page = $cgi->param("nextpage");
};
if (! exists $pages{$page}) {
  $page = "welcome";
};

# Parameter validation ommitted, as it's simply a walk of the validati
+on chain for every page ...

# Execute the special code for this page, so all this searching had so
+me sense :
&{$pages{$page}};
[download]