Dear testers: Yes, we model!

5 thoughts on “Dear testers: Yes, we model!”

1. Hi Albert,
   Can you elaborateon how the ‘test models’ relate to existing models, created by developers? And can you identity preferred types of models, from a testing perspective?

   Reply

   1. Hi Ben,
      Thank you for your interest.
      Both testers and developers need to know the functional requirements of the system in order to perform their work. Developers need to know the expected behavior of the system in order to know what to build and testers need to know the expected behavior of the system to know what to test.
      Therefore, both roles could use the same models since they need the same knowledge. In fact, the idea is reusing and sharing functional knowledge. However, in many situations, functional requirements are not modeled in any explicit and structured form during development and then testers need to “recover” these knowledge by experimentation. Even in case that functional requirements are modeled in development phases, these are not validated. Testers can validate them as they design test cases that can be used not only to test the system but also the modeled knowledge by executing test cases on the (previously created or recovered) model.
      Any kind of model aimed at specifying functional requirements can be use: From an structured textual description to models specified in a specialized/formal conceptual-oriented language (such as the Unified Modeing Language promoted by the Object Management Group). When using structured modeling languages, then automated validation and simulation techniques are applicable.

      Reply

2. Hi Albert,
   Another question: how do you cope with testing a system which has a lot of interfacing and dependencies with other systems? Are the models you use being fit for this integration? In today’s world, testing integration of systems if often more challenging than testing one (isolated) system. So we need to relate different models with each other? Who is best positioned for this?

   Reply

   1. Hi Geert,
      When developing/validating a model it is required to define its scope according to the project context. Certainly, a model may require knowledge of other dependent systems that is outside the scope of the model. If the models of external systems are not available, this external knowledge can be used in our models by mock-ups. This is the same technique used in system testing when we need to use unavailable dependent external functions that are not part of the testing scope.

      Reply

   2. Hi Geert,
      When developing/validating a model it is required to define its scope according to the project context. Certainly, a model may require knowledge of other dependent systems that is outside the scope of the model. If the models of external systems are not available, this external knowledge can be used in our models by mock-ups. This is the same technique used in system testing when we need to use unavailable dependent external functions that are not part of the testing scope.

      Reply