Carlo Contavalli

ccontavalli at masobit.net

Why should I use mod-xslt2? What can I do with mod-xslt2? Why did you spend your time on mod-xslt2? Those are some questions this document will try to answer. As many of you already know, xml is a relatively new language studied to represent any kind of information using simple text files. The nicest thing about XML is that it is used to store data, but it does not carry information about how that data should be displayed (rendered) on a generic media (like a screen, or a printer, ...). Informations about how to render XML files on a given media are carried along in Stylesheet files, either CSS version 2 or XSLT stylesheets. XSLT is a generic language which can be used to transform XML files to almost any other language or format. It is most used to convert XML files into HTML files (to be displayed on a screen) or to FOP (to be printed or converted to PDF). By having a file carrying the data and one carrying informations about how that data should be rendered on a device, you can separate the task of writing your web softwares from that of designing good and nice interfaces. This separation allows you to have a designer prepare the graphics of your internet site and a programmer write down the needed scripts/tools, improving the quality of the whole result. As you probably know, most programmers are not good at web design, and most designers are not good programmers. Although many companies are nowdays pushing the XML market by commercializing their wonderful XML support, very few products do have even a good support for it. As usual, many of them developed their own extensions or decided that the standards were not that good to follow, and made their products incompatible even with different versions of themselves. Talking about the world of ISP, the result is that most browsers do not support rendering XML pages using XSLT stylesheets, and even when they do, they perform a poor job, generating a different HTML output every time you refresh your page or every time you switch from one version to the next one of the same software, or just generating the wrong result. The idea behind mod-xslt2 is to cut away all those kind of problems and to parse xml files into HTML (or any other format) before they even get to the browser. That way, you will be sure the document will produce the same HTML on every request, and that any browser will be able to display your XML pages. The browsers accessing your web site won't have direct access to your XML data, and you will be able to protect it from those who may just decide to store your .xml pages on their databases and use them by themselves, unless you decide to allow it (now it's your choice). As you probably may guess, this site is a direct result of using mod-xslt2: its graphic was made by a designer on its spare time, without knowing anything about the php scripts I was going to write/use... None of us alone would have been able to reach such a good (I believe) result. Nowdays on the internet you can find many apache modules or web components able to parse XML into HTML. Some of them depend on a particular scripting language being used or require web developers to user their routines or their libraries for the XML to actually be parsed. Others can be thought as components of the web server and work mostly like mod-xslt2. However, many of them have lot of limits: they can't fetch XSLT from remote servers or from dynamic scripts, they can't parse the output of PHP or CGI, they only work with one particular version of a web server and none of them (I know of) is able to choose a different XSL to use depending on the conditions the XML was asked. While providing most of the standard features, mod-xslt2 has some extensions that when used correctly won't break compatibility with other xslt processors and that will allow you even a greater deal of flexibility. While working on mod-xslt2, I also wanted to have a clean API which could be used on other web server then the ones I was working on, and wanted to have a flexible and expandable infrastructure to add many new features. Most of them are still under development but I feel like many of the targets I wanted to achieve have now been reached. The API has already been used both from Apache 1.3 (stable) and Apache 2.0 (I still need a bit of testing, since I'm one of those who don't trust Apache 2.0 yet and do not run it on their production systems). Hooks are also provided to easily implement result cacheing and work is in progress to have a CGI and Proxy parser. There are even many features that are under discussion right now that will be probably added day by day. Take a look to the future direction pages to know more about those features.