first part of fix for attributes and roles mess. metclass coompat bug still lurks

[catagits/Reaction.git] / lib / Reaction / Manual / Intro.pod

=head1 NAME

Reaction::Manual::Intro - Introduction to Reaction

=head1 SYNOPSIS

  Moose + Catalyst + Layered Models = eternal happiness

=head1 MOTIVATION

While L<Catalyst> is a very powerful and flexible web app 
framework, many pieces end up being recoded for each 
new application: user interfaces, in particular.

Meanwhile, advances in Perl metaprogramming (especially L<Moose>)
have opened up new possibilities in terms of introspection.
Why not build user interfaces based on class metadata?  
This would allow the user interface templates to be very general 
where possible.

Thus, Reaction is basically an extended MVC framework, which 
leverages class metadata to reduce or eliminate UI coding.

But it's much more....

=head1 DOMAIN MODELS AND INTERFACE MODELS

Many programmers are now comfortable using ORMs of one sort or 
another - L<DBIx::Class>, L<Class::DBI>, Hibernate, etc.  These systems 
are wonderful for bridging from the world of OO into the world of relational 
databases (or other datastores).  This model is sometimes called a 
"domain model", because it models the "nouns" of a problem domain in 
the real world.  Domain models are easily shared accross applications, 
and can enforce validation and other integrity constraints.

However, over time, many application developers find themselves adding 
business logic to the domain model.  This business logic is often 
application-specific, and reduces the reusability of the domain model.
Worse, business logic becomes spread between the model and the contoller.

Reaction adds another layer, the Interface Model.  Interface models provide an 
adaptor to the domain model, customized for a particular application (or group 
of use cases).  This decouples the domain model from the application, 
allowing it to be reused more freely.  Additionally, the Interface Model 
becomes the natural location for business logic.

Happily, Reaction again uses reflection to make the degenerate case easy - when your 
IM has no customized functionality, it can simply delegate all work to the DM.  When 
you need to add custom business logic, you can add or replace functionality as needed.

The DM/IM split is sometimes referred to as a FacadeModel - see 
L<http://www.twinforces.com/tf/docs/MFCV.html>, for example.

=head1 WHAT YOU'LL NEED TO KNOW

Reaction is based on the L<Catalyst> web application framework.  You'll certainly
need to be familiar with L<Catalyst::Manual::Intro>.

Currently, only L<DBIx::Class> is supported as a domain model.  At least basic 
familiarity will be needed. L<DBIx::Class::Manual::Intro> is a good starting point.

The default view renderer is L<Template::Toolkit>.  To edit your views, you'll 
need to know something about it.

While you don't need to know L<Moose> directly, a lot of the concepts of 
metaprogramming will keep coming up as you work with Reaction.  
Thus, getting to know L<Moose> will serve you well.

=head1 NEXT STEPS

The L<Reaction::Manual::Overview> document tries to tie the parts of reaction
together to form a big picture.

If you'd like an example, see L<Reaction::Manual::Example>.

If you're ready to dive in and start learning step by step, see 
L<Reaction::Manual::Tutorial>.

As you encounter unfamiliar terms, or want to see how a particular term is used in the 
context of the Reaction project, refer to the L<Reaction::Manual::Glossary>.

=head1 SEE ALSO

=over 

=item * L<Reaction::Manual::Cookbook>

=item * L<Reaction::Manual::FAQ>

=back

=head1 AUTHORS

See L<Reaction::Class> for authors.

=head1 LICENSE

See L<Reaction::Class> for the license.

=cut