What criteria do you use to decide whether execute a test on target or on host in a sprint? We have integration testing using a server for an embedded system project.

We are starting to learn and researching on how to apply integration testing and I do not know how exactly decide. What are the scenarios for each approach?

I suppose that when is needed to test interaction in a few no critical modules I can do it on host. But when they are critical, and the timing is important it is better to do it on target.


It depends on:

  • the cost involved in uploading executable to target, controlling said target and downloading test outcomes. If the cost is high, the test is less frequently run.

  • the physical environment available on target but not on host and needed for test. For instance a target-specific chip to implement Wifi.

  • the availability of target itself. Sometimes hardware and software are designed in parallel. Or the actual target is too expensive to be left to random testers (like space critical computers).

  • Does your target need human interaction when you run your tests? I am supposing the automation would not be affected by the target interaction. – PySerial Killer Oct 27 '17 at 14:57
  • Some targets do, others don't. Sometimes a test goes really wrong and tester needs to manually reset the board. – mouviciel Oct 27 '17 at 15:09

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.