EXAMPLE 4.1

04 May 2007

State of Utah

Department of Human Services

Division of Child and Family Services

SAFE Project

Implementation and Use of IBM Rational Robot Automated Testing Tool

There are many factors involved in the successful implementation of an automated testing tool on a project. Some of the most important are having enough resources to create and execute the scripts, expert knowledge of the tool, and a stable application under test.

Due to there being only one full-time resource available for automated script creation on the SAFE project, we have not attempted full end-to-end script coverage. Rather, we have built scripts that provide the most value. Currently, we have scripts that cover the following areas:

1) Initial build testing (Smoke Test)

2) Main functionality testing of existing areas of SAFE (Regression)

3) User Acceptance testing of main areas of SAFE (UAT)

4) Test data creation

5) Training data creation

The scripts for the initial build testing go through the application opening windows and verifying menu items and dropdowns, providing a quick, high level stamp of approval that the build is suitable to use for more thorough testing.

The scripts for the main functionality testing of existing areas of SAFE cover referral and case creation of various case types and go through the basic processes involved to take the cases through closure.

The scripts for the user acceptance test of main areas of SAFE are currently very limited and, for the most part, only cover referral and CPS/CFA case creation right now. These are being expanded as time permits.

The scripts for test data creation utilize datapools to insert large amounts of test data into any database. We found there was a need for developers and testers to have “personal” test data available so these scripts were created to insert data that is personally identifiable for testing purposes. Additionally, one tester has specific needs for certain test scenarios so scripts were created to actually insert all necessary people and create cases to accommodate the tester’s needs. As time permits, we will create scripts that will actually automate what the tester does to verify each test case and scenario.

The scripts for training data creation utilize datapools to insert the same test data over and over with slightly changed names. The training team had a need to do multiple trainings at the same time, and, therefore, they needed more than one set of training data. We decided to create scripts that would loop through the same code entering data identifiable to each region of Utah that will be receiving the training each time through the loop. This is one of the rare times where it made sense to spend the time and money to create automated scripts that would only be used one time because the same test data needed to be entered many times and it also greatly decreased the likelihood of mistakes being made during data insertion as there would be entering the data manually. Also, though these scripts were only planned to be used one time, we have already found the need to execute them on another database and they can be modified to fit additional data insertion needs down the road.

As we go forward, the main focus of automated testing on the SAFE project will be to keep existing scripts up-to-date as new functionality is implemented and to broaden the coverage of the existing regression and UAT scripts. Additionally, we will continue to serve the developers and testers as they come up with needs that are deemed worthy of automating.