The direction of development application on Internet is toward the Web. To add interactively to web pages, many different techniques can be used such as CGI, HTML with scripting language, ActiveX controls, Java applet, etc.

Dynamic HTML, a new HTML specification released by Microsoft and the World Wide Web Consortium (W3C), provides the ability to manipulate any page element, change element styles and colors, and position elements dynamically, that is, at runtime (Cluts, 1998). This will help reduce the load on Web server�s resources.

Scripting language is good for some jobs that do not require a full Web program running on the server, nor do they require the power of ActiveX or Java. But its solution can only function to the level that the languages support (Mckay, 1998).

Scripting language is generally used for client-side programming. The advantages of the client-side programming are (McComb, 1997):

• It provides a means to create dynamic HTML pages without requiring programming access to the server.
• No additional server interaction is required. This reduces the load on the server.
• Generally client-scripting language is easier to write than a program design to run on a Web server.

There are several new scripting languages. Two widely used scripting language in existence today are JavaScript and VBScript. Both JavaScript and VBScript can also be used for server-side programs, but this use is not nearly as common as client-side programs.

JavaScript is developed by Netscape. JavaScript is an easy to learn scripting language designed to enhance the capabilities of HTML. There is also an extensive use of JavaScript. Netscape provides LireWire for managing Web sites and creating server software (It includes JavaScript Compiler). Netscape server can run compiled JavaScript application.

VBScript is a Microsoft�s Web scripting language. It can be used to integrate Java applets, ActiveX controls, intrinsic HTML form controls, and HTML to build dynamic client side applications for the Internet and can also tie other applications that support OLE technology (Breedlove, et al., 1996).

If we compare JavaScript and VBScript. JavaScript is supported by all three share-market browser an the applications can be reached by a wider audience.

Java is another form of client-side programming, but its capabilities and complexities take it beyond JavaScript and VBScript. Java is an object-oriented programming language, developed by JavaSoft. Its syntax is similar to C and C++. Java code can be compiled into a stand-alone executable or can be compiled into a machine-independent format consisting of virtual machine instructions and symbolic data, called the bytecode format. With the characteristic of this format, Java software is cross-platform and can run on any system with a Java Virtual machine(VM)(Hamilton, 1996). Each Java VM is written for a specific platform, and translates the more generic Java instructions into platform.

Java�s one special kind of Java program known as applet can be attached to a Web page�s HTML document and interpreted by Java-enabled browsers such as Netscape Navigator 2.0x, 3.0, Microsoft Internet Explorer (McGraw et al., 1996). Such Java-enabled browsers include complete Java VM. It automatically downloads and begins running any Java applet it finds embedded in a Web page. One benefit of using applet is that it can run on any platform that has a Java VM and also it has likely wide availability of re-usable Java Applets.

When performance does matter, there�s no denying that interpreted Java byte code is slow. Developers who need maximum performance have a good reason for avoiding Java. Compiled program like C or C++ that runs at least 10 times faster than an interpreted Java program is a better choice (Halfhill, 1997).

A comparison of JavaScript, VBScript and Java (McComb, 1997;Chandak, 1997)

ActiveX control is a Microsoft technology by redefining OLE technology for the Internet(Chandak, 1997). A set of technologies called ActiveX consist of ActiveX controls, ActiveX scripting, Active Docs, Active Server Page. ActiveX controls are interactive objects that can be created and embedded into any application including Web and window application. ActiveX controls are great for accessing system services, threading, providing user interface components(Cluts, 1998). One advantage of controls is that it can be speedy, because developer can use powerful languages, such as C or C++ to create them. In the web page that includes ActiveX control, the browser will automatically (actually depending on browser�s setting ) download and install, if the control does not already exist on the user�s system.

ActiveX can be viewed and executed by all window platforms and internet Explorer, while Netscape Navigator has to use a third-party plug-in-ScriptActive from Ncompass Labs.

There are four primary methods of running programs on a Web server :CGI, FastCGI, API and SSI.

CGI is a mechanism that allows Web servers to execute special-purpose CGI programs to handle specific requests from a Web browser (Evans and Rogers, 1997). A user interface on client can be defined using the HTML FORM tag. User can request some operation on the input by "submitting" the form.

Advantages of CGI (Liu, 1998; Evans and Rogers,1997; Ju, 1997):

• The protocol is simple. An experienced programmer could understand CGI fast.
• The CGI program can be written in most languages. For example: C/C++, Perl, TCL, UNIX shell, etc.
• Nearly all Web servers support it.
• CGI programs run in a different process separate from that of the Web server. The benefits of this separation are stability and security for the server. The Web server is unaffected by the robustness of the CGI programs that may occur.

Disadvantages of CGI (Liu,1998; Evans and Rogers, 1997):

• Developers have limited control over UI component layout (HTML tag).
• Remote operation arguments are strings only and have no structured types.
• The Web server spawns a new CGI program instance to handle each invocation.
• Each remote invocation is slowed by acquisition of needed resources.
• Its performance doesn�t meet the needs of serious web applications, mainly because a CGI process exits after servicing a request. So each browser request includes process startup and exit overhead.
• Since CGI program is run on host computer, the possibility of the abuse by CGI programs should be considered.

The fact that Web server forks a new process whenever a CGI script is run makes CGI proved to be slow for sites receiving thousands of hits per day. FastCGI is a performance-enhanced version of CGI, which allows the Web server to run a CGI as a single process rather than generate a new process for every CGI request. FastCGI is actually CGI with only a few extensions (Ju, 1997):

• Like CGI, FastCGI is also language-independent because they are simply protocols. A protocol can be implemented in numerous ways under any language.
• Like CGI, FastCGI runs applications in processes isolated from the core Web server, which provides greater security and robustness than APIs
• Like CGI, FastCGI is being positioned as an open standard. Free FastCGI application libraries and upgrade modules for popular free servers are avaliable.
• Like CGI, FastCGI is not tied to a particular Web server architecture and is therefore stable even when server architecture changes.
• Distributed computing. The FastCGI application can be run on a different machine from the Web server. Distributed computing is a proven technique for scaling, load balancing, and improving system availability.
• Multiple and Extensible roles. CGI can be considered to be operating purely in a responder role. FastCGI application can do that and more, such as perform modular authentication and authorization checks and translate data from one type to another. As new roles are needed, more definitions and FastCGI programs can be written to fill them.
• Sharing Memory with other processes. With FastCGI, different instances of the same application can access the same file from the same section of memory, resulting in the swift retrieval of data from memory as opposed to disk.
• Allocated Processes. Like applications written with Web server APIs, FastCGI applications do not require the Web server to start a new process for each application instance.

The main disadvantage of CGI is performance. CGI requires processes to be spawned whenever a request is made. This may consume additional CPU resources, time, and memory required to load, for example, a Perl interpreter or dynamically linked C libraries. Web server APIs provide a way for concurrent instances of the CGI program to share common resources and eliminate the repetitious overhead costs.

API programs are designed to run under a given server, such as the Netscape Commerce Server or Microsoft�s IIS server, and are based on a set of programming commands that can be accessed using a compatible programming language (McComb, 1997). The API for Netscape servers is known as NSAPI and the API for Microsoft servers is known as MSAPI..

Advantages of API (McComb, 1997; Ju, 1997):

• A server API program is generally faster than a CGI-only version since server API programs run as dynamically loaded libraries or modules instead of separate executables.
• The difference between a CGI and a server API program is in the way the common resources are shared. When a CGI program is requested, the Web server spawns a process. A running CGI process has its own instance of executable code and its own set of data. For API program, additional instances of the API program are implemented as additional threads, and common resources (data or functional code) can be shared among the processes, and between the process and the Web server.

• Server API programs are closely tied to the server they work with. Changing to a different Web server, the program must be rewritten.
• Server API programs are as dependent on platform and programming language as is the Web server.

Contents

Next

---

	Designed by Chollada Wiyaporn.
	Copyright � 1998. All Rights Reserved