The CGI.pm module makes it much easier to produce HTML code from CGI scripts written in Perl. If your goal is to produce self-contained CGI applications that include both the program logic and the interface (HTML), then CGI.pm is certainly the best tool for this. It excels for distributable applications because you do not need to distribute separate HTML files, and it's easy for developers to follow when reading through code. For this reason, we use it in the majority of the examples in this book. However, in some circumstances, there are good reasons for separating the interface from the program logic. In these circumstances, templates may be a better solution.
HTML design and CGI development involve very different skill sets. Good HTML design is typically done by artists or designers in collaboration with marketing folks and people skilled in interface design. CGI development may also involve input from others, but it is very technical in nature. Therefore, CGI developers are often not responsible for creating the interface to their applications. In fact, sometimes they are given non-functional prototypes and asked to provide the logic to drive it. In this scenario, the HTML is already available and translating it into code involves extra work.
Additionally, CGI applications rarely remain static; they require maintenance. Inevitably, bugs are found and fixed, new features are added, the wording is changed, or the site is redesigned with a new color scheme. These changes can involve either the program logic or the interface, but interface changes are often the most common and the most time consuming. Making specific changes to an existing HTML file is generally easier than modifying a CGI script, and many organizations have more people who understand HTML than who understand Perl.
There are many different ways to use HTML templates, and it is very common for web developers to create their own custom solutions. However, the many various solutions can be grouped into a few different approaches. In this chapter, we'll explore each approach by looking at the most powerful and popular solutions for each.
One thing we won't do in this chapter is present a novel template parser or explain how to write your own. The reason is that there are already too many good solutions to warrant this. Of the many web developers out there who have created their own proprietary systems for handling templates, most turn to something else after some time. In fact, one of your authors has experience doing just this.
In both of my attempts, I realized the solutions were not worth the effort required to maintain them. In the second case, the code was very maintainable, but even minor maintenance did not seem worth the effort given the high-quality, open source alternatives that are already tested, maintained, and available for all to use. More importantly, custom solutions require other developers and HTML authors to invest time learning systems that they would never encounter elsewhere. No one told me I had to choose a standard solution over a proprietary one, but I discovered the advantages on my own. Sometimes ego must yield to practicality.
So consider the options that are already available and avoid the urge to reinvent the wheel. If you need a particular feature that is not available in another package, consider extending an existing open source solution and give your code back if you think it will benefit others. Of course, in the end what you do is up to you, and you may have a good reason for creating your own solution. You could even point out that none of the solutions presented in this chapter would exist if a few people hadn't decided they should create their own respective solutions, maintain and extend them, and make them available to others.
Copyright © 2001 O'Reilly & Associates. All rights reserved.