One of the main challenges of test automation is the maintenance of the different scenarios. In an article, the publisher Ranorex reviews best practices for creating tests that are easy to maintain and optimize.
Designing easily maintainable tests is played out at several levels: it starts at the design phase and continues when writing the test cases. Even in the execution phase, certain practices make it possible to optimize test management in order to facilitate maintenance.
Best practices when designing tests:
Decide what to test before deciding how to test it before determining which tests should be performed manually and which should be automated, it is first necessary to identify all the functionalities to be tested, then for each of them establish sequences to be tested with a clear description of the expected results.
Document test cases:
This documentation step makes it possible to verify that all the necessary elements are provided for each test: preconditions, test sequence, description of the expected behavior.
keep it simple:
Ideally, each test should be limited to a single feature and only fail one case. The more complex a test, the more likely it is to have flaws and flaws that will then need to be corrected.
Use naming standards:
The names of interface components and test objects must be sufficiently self-explanatory. If comments are needed, this is a sign that the test may be too complex.
Best practices when writing tests:
-Use a modular structure:
A test case must be able to be executed separately from the others. Care must be taken to limit dependencies between different test cases as much as possible. If it is nevertheless necessary, the test framework must make it possible to ensure that the test cases are always executed in the correct order.
-Create tests loosely coupled to GUIs:
To prevent a change in the interface from forcing a test to be rewritten, the latter must be coded in such a way as to make it as independent as possible of the UI. Better to use unique IDs for each GUI component rather than coordinates or hard-coded HTML elements.
-Group tests by functional area:
This makes it easier to update the tests if a change occurs in this area.
-Avoid copying and pasting code:
It is better to create reusable modules than to copy the same code in several test cases. If a change occurs on a recurring sequence, it will suffice to modify the associated module rather than review all the test cases.
-Separate test steps from data
In general, in testing as in development, hard-coding data should be avoided. The use of variables and parameters makes maintenance much easier.
-Use versioning tools:
These tools facilitate collaboration and allow you to revert to an earlier version of the code in the event of a problem. Ranorex recommends using the same tools to manage application code and associated test cases.
Best practices when running tests:
Check that the test environment is stable.
Unreliable servers or connections can cause otherwise perfectly stable tests to fail.
Use initiating and closing processes:
A launch process (setup) ensures that the environment meets all the conditions required for the proper conduct of the tests (application launched, a user connected, etc.) A closing process restores the environment to its state initial, including cleaning up test data.
Stop quickly in the event of a major malfunction:
In the event of a serious problem (lack of response from the application for example), it is useless to continue the test. It is better to escalate the problem immediately rather than wait.
Stop the test only if necessary:
Interrupting a test at the slightest isolated error wastes time. Sometimes, it is better to roll out all the test cases, in order to check if they are going well despite everything.
Separate the tests which aim to elucidate a defect from the others:
It is best to run tests that address unresolved defects or less robust tests separately from the rest. Thus, if the main tests have abnormal results, it will be easier to be sure.