Jan Kester Blog

Core principles of good tests When you start automating your tests, there a few important principles to stick to. Reliable First your tests must be reliable. This means that they must repeatable and always give the same results. They should not contain any environment dependencies, and be executable on any environment. To get this working, all hard-coded settings must be configurable, a maven or ant script is required to build and configure the test environment, and all code, test date and configuration scripts must be checked in source code repository. Having reliable tests that run everywhere and always, also means that your tests must take care to setup necessary test data, and clean it afterward again. Fast Next, it would be nice when your tests are fast. In the lifetime of a project you will build up ever more tests; with every next sprint you will need to run more tests. You can't run all your tests on check-in, and for certain groups of tests need to schedule test exe

Test automation importance With our change to agile, we have made an important paradigm shift. In the old days we would stick to our planned roadmap, and compromise either on delivery dates or software quality. In Agile software development we are willing to compromise on scope, but not on software quality or delivery dates. This means in short: no code goes out untested.   What tests to execute? It is nearly impossible to test everything. There are so many types of tests, like: regression, story acceptance, developer, integration, usability, performance, security, upgrade, user acceptance etc. In our release planning, we must make a conscious decision which of those types we want to execute. Some of the test types may require additional experts, some of the tests may not be relevant, some may be just too costly.   The usual scrum tests Within our sprints, we tend to stick to regression tests, story acceptance tests, developer tests and integration tests. Developer tests ar