Making use of a PC based tool has a number of advantages over using Host based testing software, and the most significant advantage is usually that of ease of use. Another key benefit is typically host platform independence.

Most Host based testing software is written by programmers for programmers, and usually requires considerable programming skills to use effectively. Some PC based application testing software also suffer from this problem requiring extensive programming skills, but testing tools are available which can be used successfully by people with no programming training whatsoever. These tools make use of the PC interface to provide simple and easy to learn testing functionality.

"If a PC based application testing tool is to be successfully used to test Host based application software, then it must have been designed specifically to test host based applications."

This may seem fundamental but selecting a tool which has only been designed to test PC based applications (whether that testing be in DOS or Windows) and expecting to use it successfully to test Host based applications is a recipe for problems. It will inevitably result in either disappointment or significantly increased work in setting up and maintaining test cases.

There are three key functional tool requirements for minimising the effort to test host based applications, which some PC based tools cater for but many don’t.

• Auto host pacing

• Suspension of processing dependent on host processing requirements

• Processing screen input and output based on text not bit maps

The most important requirement for host based testing is that the tool be designed to automatically handle variable Host response times.

In addition to using the ENTER and TAB keys, a typical Host application may use as many as 15 or 20 program or special function keys. Each of these keys can result in a delay while the Host performs some processing action. This delay may be quite variable depending on the function to be performed, and on the load existing on the Host at the time that the test is run.

Some PC based application testing software attempts to support application software response delays by re running the test at exactly the same pace at which it was recorded. In a straight PC environment, where typically only the single task will be executing, this method of playback pacing is usually quite adequate. However, in a Host based environment, this method is virtually useless.

The application testing tools need to be able to recognise when something happens that may result in a delay from the Host application, and having recognised that such a variable delay may exist, be also able to recognise when the Host has completed its processing and is ready to accept further input from the testing workstation. As an example TALC2000 supports this form of automatic Host pacing, and will adapt to provide the maximum playback speed which the Host is capable of handling based on its current work load.

Quite commonly, a Host application that has begun a lengthy data base search will display a message at the user's terminal indicating that the search is in progress. While a message such as "Searching, please wait..." is displayed at the terminal, the test playback tool must pause, and resume playback when the message disappears. Again, depending on Host performance this message may be displayed for greatly differing times during different test playback runs. As an example TALC2000 provides an easily used method of monitoring for these messages, and suspending operation for what ever time is required.

A typical Host based application, even if it is running using a standard Windows terminal emulator, displays data in character mode, as if that data was displayed on a "dumb" terminal. Depending on the tool it will treat the display as text based character display or a windows bit map.

As an example TALC2000 supports a text based character display and provides a "text mode" screen capture function which allows Host responses to be monitored in the form of an internally created character buffer. By using this method of screen capture TALC2000 can compress a typical Host response screen into between 700 and 900 bytes of information, as distinct from the 20,000 to 30,000 bytes of information which would be necessary to store the same data as a compressed Windows Bitmap.

This not only results in a significantly lower demand for disk space for storage of test information (typically better than a 40 times reduction), but also greatly increases the speed of test validation.