We wanted to start the new-year off with a topic we hear a lot about from many, many people.
A couple of weeks ago we talked about the issue of Software Testing being Ad Hoc and some of the reasons this might be the case. See the blog from January 14, 2019.
The major concerns were loss of information, repetition and reinvention.
The quickest way to stop this loss is to classify the testcases. This does not have to be complex at the beginning. You can classify the testcases based on the following criteria:
- Priority – High – must be executed with every release; Medium – must be executed at least once a release; Low – may be ignored unless there is a lot of time.
- Feature/Module/Function – This does require some thought at the beginning since you want to get the classification right to avoid having to do a lot of reclassification later.
- Level of testing – Unit, Integration, System, Acceptance – again something that might require some thought at the beginning to avoid having to reclassify things later.
Obviously this is best done in a tool with a database behind it although a lot of companies start with a spreadsheet. The only expectation is that the next iteration of the project can at least see what exists and pick up what they need.
Later you may determine other classifications that are useful. If you do, you may or may not want to reclassify everything that has already been stored. You may also want to do it over time as the testcases are used rather than trying to do it as one project.
We have gone one step towards eliminating the ad hoc nature of the testing and now have something that can be reused.