Web Development Technologies

Ragu Bharadwaj, Daylight CIS Inc.

I would like to describe a methodology we are beginning to use at Daylight for the development of web applications. The key advantages are total separation of information and presentation. This allows our users to vary the presentation of any compiled web app that we ship. The method uses 2 technologies which I will describe.

XML (eXtensible Markup Language)

A way of representing information in a tree-based (or container-based) scheme. It is similar to how we do so in a TDT (Thor Data Tree), with the difference that TDTs can have only a couple of levels of children. In XML this restriction is removed.

An Example:

<label>
    <name> John Doe </name>
    <address>
        <street> 123 Main Street </street>
        <city> AnyTown </city>
        <state> NM </state>
        <zipcode> 87501 </zipcode>
    </address>
    <phone> 505-989-1000 </phone>
</label>

__________________________________________________
| <label>                                        |
|    ____________________________________________|
|    | <name> John Doe </name>                   |
|                                                |
|    ____________________________________________|
|    | <address>                                 |
|    |     <street> 123 Main Street </street>    |
|    |     <city> AnyTown </city>                |
|    |     <state> NM </state>                   |
|    |     <zipcode> 87501 </zipcode>            |
|    | </address>                                |
|                                                |
|    ____________________________________________|
|    | <phone> 505-989-1000 </phone>             |
|                                                |
| </label>                                       |

So what's the big deal:

More and more applications are using XML as a format they can read from and write to. The XML of course has to obey certain rules (for example, each label tag must have a name tag or the phone tag must only contain numbers) These rules are encoded in DTDs or Document Type Definitions.
What this means is that data transfer between applications is highly eased with XML.
XML is a simple, plain text format very similar to HTML. This makes debugging code that uses XML very easy.
Several open source libraries and tools are available for parsing XML documents. These tools are very well debugged and optimized both for speed & for ease of use. Using these tools with XML greatly improves productivity. Almost every computer language(C++, Perl, Python, Ruby, Java) has such tools in the open source.

XSLT (eXtensible Stylesheet Language: Transformations)

Given an XML tree, such as say

<label>
    <name> John Doe </name>
    <address>
        <street> 123, Main Street </street>
        <city> AnyTown </city>
        <state> NM </state>
        <zipcode> 87501 </zipcode>
    </address>
    <phone> 505-989-1000 </phone>
</label>

typically, one may want to convert it into a printable address label like:

John Doe,
123, Main Street
Anytown, NM 87501

followed perhaps by a bar code for the same

To do so one could

place rules for doing the same in an XSLT file
run a program that reads in the XML file for the label and the XSLT file for conversion
apply the conversions read to the XML file read in and produce the output

Advantages:

XSLT is simple & easy to learn. As HTML is related to XML, XSLT has special constructs for outputting HTML
A web CGI/app can be delivered in a compiled form with XSLT files that are plain text. This enables the users to alter the XSLT file independently, and alter the appearance of the output page without any recompilation (which would have needed source code)
Depending on the invocation context of a CGI/web app, the output can be

HTML for a web browser
WML for a PDA based platform such as PalmOS
PDF
an XML format that can be directly imported into Microsoft Office

Web App Design

We plan to use the following scheme

DB -> XML -> XSLT -> HTML, PDF, WML,...

A Web Application will typically

query and retrieve information from a database
format the retrieved information in generic XML
invoke a XSLT parser library, pass it the name of an XSLT file to read and the XML created
the XSLT parser will then read in the XSLT from the XSLT file and
transform the passed in XML using the rules mentioned in the XSLT file
output the transformed data

This allows for:

content & presentation separation

all content is in XML and is retrieved by the web application. Content is created by a compiled program, optimized for speedy retrieval. For example, the web application can use pooled Db connections to talk to a database
all presentation information is in XSLT. XSLT files are plain text and can be modified by the end user or administrator. This enables tailoring output according to one's custom requirements. For example,

the corporate logo, with a search date can be inserted into the output.
certain irrelevant fields can be deleted from the output.
The output can be made to be Microsoft Office compatible
The output can be augmented with additional info from other programs

Example apps:

Savant: Spresi Literature Search
ReacsSavant: Chemsynth Reaction Database Literature Search

There are several other web technologies that Daylight may offer solutions for in the near future.

Taglibs in Java (via JSP taglibs), PHP, Ruby & other languages
Web Application aRchives (WARs) which are java-based web apps that can be deployed in any standard Application Server such as BEA Weblogic or IBM WebSphere.