What is regression testing?
Regression testing is re-testing of a previously tested program following modification at any stage of the software development cycle. This ensures faults have not been introduced or identified as a result of the changes made. The outputs from the original software will have been checked so that in regression tests it is possible to check that the same outputs are achieved.
How can the LDRA tool suite assist with regression testing?
- With applications which may require test outputs to be identical to the previous test run, the entire regression testing activity can be fully automated with LDRA Testbed® and TBrun®, LDRA's test harness generator.
- LDRA Testbed can be integrated with other tools such as capture/playback systems. Integration of LDRA Testbed and other software products is possible using the LDRA Testbed Standard Interface File (SIF).
When would you use it?
Regression tests are designed for repeatability, and are used when testing updated versions of the system. Regression testing may be performed on many different application types, including embedded software, e-Commerce and web-based systems. The facilities that are provided by the LDRA tool suite enable the regression testing process to be applied from the unit level up to and including system integration.
How can the regression testing process be optimised?
When a number of test data sets are used to test a software application, coverage is accumulated from each test data set, leading to an accumulated coverage total. With a large set of test data, however, data sets may ‘overlap’ in that some may not contribute to the accumulated coverage total in the sense that the areas of the software application that they exercise are also exercised by other, more extensive, data sets. Data sets that overlap in this manner may be said to be redundant.
The LDRA tool suite provides a Data Set Analysis module which analyses all of the test data sets used during the test process and reports the smallest set which will achieve the equivalent accumulated coverage total. In so doing, the tool suite clearly identifies all of the redundant data sets which may then be removed prior to regression testing. Removal of the redundant test data sets leads to an increase in speed and efficiency.
What results will be produced?
Annotations can be used to detect changes in metrics. The user records the value of specific metrics in the source code using comments. LDRA Testbed expands the comments and compares the predicted values with the actual values achieved with the current analysis run. The use of annotations can significantly reduce the costs of regression testing since it reduces the time engineers spend looking at detailed test reports.
|
