http://qs1969.pair.com?node_id=612534

wfsp has asked for the wisdom of the Perl Monks concerning the following question:

I'm working on a CGI::App that is growing like topsy. The list of runmodes has past a dozen and is veering crazily toward two.

I remembered dragonchilds comment "every major subsystem... (would) have their own C::A". He posted some code to show how that might be done

I hit some snags implementing it, here is what I finaly came up with.

Main.pm

package App::Main; use strict; use warnings; use base qw{CGI::Application}; use App::One; use App::Two; sub setup { my ($self) = @_; $self->start_mode('one'); $self->mode_param(path_info => 1); $self->run_modes([qw{ one two }] ); } sub one { my ($self) = @_; my $one = App::One->new; $one->run; exit; #$self->header_type('none'); } sub two { my ($self) = @_; my $two = App::Two->new; $two->run; exit; #$self->header_type('none'); } 1;
One.pm
package App::One; use strict; use warnings; use base qw{CGI::Application}; sub cgiapp_init { my $self = shift; $self->tmpl_path(q{tmpl}); } sub setup { my ($self) = @_; $self->mode_param(path_info => 2); $self->start_mode('one_a'); $self->run_modes([qw( one_a one_b )]); } sub one_a{ my ($self) = @_; my $tmpl = $self->load_tmpl; return $tmpl->output; } sub one_b{ my ($self) = @_; my $tmpl = $self->load_tmpl; return $tmpl->output; } 1;
Two.pm would be along similar lines. The instance script would be something like
#!C:/Perl/bin/perl.exe use strict; use warnings; use lib qw{ /www/local/sw/admin/lib }; use App::Main; my $top = App::Main->new; $top->run;
The Main.pm runmodes create new C::A objects. I've used the $self->mode_param(path_info => n); (where n tells C::A which part of path_info to use) to establish which runmode to use. So

index.cgi/one/one_a
will get us to 'main' runmode 'one', 'sub runmode' 'one_a'.

I'll now be able to develope/test each 'main runmode' separately and things should go more smoothly (I said 'should'!).

One point though, in Main.pm I've had to include exit; at the end of each method. Because of the way C::A works the header will have already been added/sent. Just having a return would mean another header would be added. Interestingly, $self->header('none') did what it says on the tin but tacked on the word 'none' to the end of the ouput (after, in this case, the closing html tag).

What do the monks think about the general approach? Worth pursuing or are there pitfalls and traps ahead?