Catalyst::Manual::About - Basic explanation of Catalyst


Catalyst::Manual::About - Basic explanation of Catalyst


This document is a basic introduction to the why of Catalyst. It does not teach you how to write Catalyst applications; for an introduction to that please see the Catalyst::Manual::Intro manpage. Rather, it explains the basics of what Catalyst is typically used for, and why you might want to use Catalyst to build your applications.

What is Catalyst? The short summary

Catalyst is a web application framework. This means that you use it to help build applications that run on the web, or that run using protocols used for the web. Catalyst is designed to make it easy to manage the various tasks you need to do to run an application on the web, either by doing them itself, or by letting you ``plug in'' existing Perl modules that do what you need. There are a number of things you typically do with a web application. For example:

Catalyst makes it easy to do all of these tasks, and many more. It is extremely flexible in terms of what it allows you to do, and very fast. It has a very large number of ``plugins'' that interact with existing Perl modules so that you can easily use them from within your application.

What isn't Catalyst?

Catalyst is not an out-of-the-box solution that allows you to set up a complete working e-commerce application in ten minutes. (There are, however, several systems built on top of Catalyst that can get you very close to a working app.)

Catalyst is not designed for end users, but for working programmers.

Web programming: The Olden Days

Perl has long been favored for web applications. There are a wide variety of ways to use Perl on the web, and things have changed over time. It's possible to handle everything with very raw Perl code:

    print "Content-type: text/html\n\n<center><h1>Hello

for example, or

    my @query_elements = split(/&/, $ENV{'QUERY_STRING'});
    foreach my $element (@query_elements) {
        my ($name, $value) = split(/=/, $element);
        # do something with your parameters, or kill yourself
        # in frustration for having to program like this

Much better than this is to use Lincoln Stein's great the CGI manpage module, which smoothly handles a wide variety of common tasks--parameter parsing, generating form elements from Perl data structures, printing http headers, escaping text, and very many more, all with your choice of functional or object-oriented style. While the CGI manpage was revolutionary and is still widely used, it has various drawbacks that make it unsuitable for larger applications: it is slow; your code with it generally combines application logic and display code; and it makes it very difficult to handle larger applications with complicated control flow.

A variety of frameworks followed, of which the most widely used is probably the CGI::Application manpage, which encourages the development of modular code, with easy-to-understand control-flow handling, the use of plugins and templating systems, and the like. Other systems include AxKit, which is designed for use with XML running under mod_perl, and Maypole--upon which Catalyst was originally based--designed for the easy development of powerful web databases. Is it not the purpose of this document to criticize or even briefly evaluate these other frameworks; they may be useful for you and if so we encourage you to give them a try.

The MVC pattern

MVC, or Model-View-Controller, is a model currently favored for web applications. This design pattern is originally from the Smalltalk programming language. The basic idea is that the three main areas of an application--handling application flow (Controller), processing information (Model), and outputting the results (View)--are kept separate, so that it is possible to change or replace any one without affecting the others, and so that if you're interested in one particular aspect, you know where to find it.

Discussions of MVC often degenerate into nitpicky arguments about the history of the pattern, and exactly what ``usually'' or ``should'' go into the Controller or the Model. We have no interest in joining such a debate. In any case, Catalyst does not enforce any particular setup; you are free to put any sort of code in any part of your application, and this discussion, along with others elsewhere in the Catalyst documentation, are only suggestions based on what we think works well. In most Catalyst applications, each branch of MVC will be made of up of several Perl modules that can handle different needs in your application.

The purpose of the Model is to access and modify data. Typically the Model will interact with a relational database, but it's also common to use other data sources, such as the Plucene search engine or an LDAP server.

The purpose of the View is to present data to the user. Typical Views use a templating module to generate HTML code, using Template Toolkit, Mason, the HTML::Template manpage, or the like, but it's also possible to generate PDF output, send e-mail, etc., from a View. In Catalyst applications the View is usually a small module, just gluing some other module into Catalyst; the display logic is written within the template itself.

The Controller is Catalyst itself. When a request is made to Catalyst, it will be received by one of your Controller modules; this module will figure out what the user is trying to do, gather the necessary data from a Model, and send it to a View for display.

A simple example

The general idea is that you should be able to change things around without affecting the rest of your application. Let's look at a very simple example (keeping in mind that there are many ways of doing this, and what we're discussing is one possible way, not the only way). Suppose you have a record to display. It doesn't matter if it's a catalog entry, a library book, a music CD, a personnel record, or anything else, but let's pretend it's a catalog entry. A user is given a URL such as Now what?

First, Catalyst figures out that you're using the ``catalog'' Controller (how Catalyst figures this out is entirely up to you; URL dispatching is extremely flexible in Catalyst). Then Catalyst determines that you want to use a display method in your ``catalog'' Controller. (There could be other display methods in other Controllers, too.) Somewhere in this process, it's possible that you'll have authentication and authorization routines to make sure that the user is registered and is allowed to display a record. The Controller's display method will then extract ``2782'' as the record you want to retrieve, and make a request to a Model for that record. The Controller will then look at what the Model returns: if there's no record, the Controller will ask the View to display an error message, otherwise it will hand the View the record and ask the View to display it. In either case, the View will then generate an HTML page, which Catalyst will send to the user's browser, using whatever web server you've configured.

How does this help you?

In many ways. Suppose you have a small catalog now, and you're using a lightweight database such as SQLite, or maybe just a text file. But eventually your site grows, and you need to upgrade to something more powerful--MySQL or Postgres, or even Oracle or DB2. If your Model is separate, you only have to change one thing, the Model; your Controller can expect that if it issues a query to the Model, it will get the right kind of result back.

What about the View? The idea is that your template is concerned almost entirely with display, so that you can hand it off to a designer who doesn't have to worry about how to write code. If you get all the data in the Controller and then pass it to the View, the template isn't responsible for any kind of data processing. And if you want to change your output, it's simple: just write a new View. If your Controller is already getting the data you need, you can pass it in the same way, and whether you display the results to a web browser, generate a PDF, or e-mail the results back to the user, the Controller hardly changes at all--it's up to the View.

And throughout the whole process, most of the tools you need are either part of Catalyst (the parameter-processing routines that extract ``2782'' from the URL, for example) or are easily plugged into it (the authentication routines, or the plugins for using Template Toolkit as your View).

Now, Catalyst doesn't enforce very much at all. Template Toolkit is a very powerful templating system, and you can connect to a database, issue queries, and act on them from within a TT-based View, if you want. You can handle paging (i.e. retrieving only a portion of the total records possible) in your Controller or your Model. In the above example, your Controller looked at the query result, determining whether to ask the View for a no-result error message, or for a result display; but it's perfectly possible to hand your query result directly to the View, and let your template decide what to do. It's up to you; Catalyst doesn't enforce anything.

In some cases there might be very good reasons to do things a certain way (issuing database queries from a template defeats the whole purpose of separation-of-concerns, and will drive your designer crazy), while in others it's just a matter of personal preference (perhaps your template, rather than your Controller, is the better place to decide what to display if you get an empty result). Catalyst just gives you the tools.


Jesse Sheidlower,


Catalyst, the Catalyst::Manual::Intro manpage


This program is free software, you can redistribute it and/or modify it under the same terms as Perl itself.

 Catalyst::Manual::About - Basic explanation of Catalyst