Tag Info

New answers tagged

1

You may want to consider only running tests that have recently failed. Given that the entire set takes 6-7 hours (i.e. is runnable nightly), you could base your tests set on the past few nights' results. If you mix in a few randomly selected tests for broader coverage over the course of the day, you should get a good chance of catching errors. This idea ...


1

TL;DR: Create a Dependency DAG from affected modules Identifying the tests that are impacted by a particular change is the same as identifying when to recompile/relink a object file. Create a dependency directed acyclic graph (DAG) starting at the modified module. You should be able to traverse all imports to identify the what needs to be tested. You can ...


7

There are certain static analysis tools that can help determine "test impact", which can then run the effected tests. But I can't help but feel that you're solving the symptom, not the problem. When I worked in QA, there was one overriding mantra that has helped me as a developer: "don't trust the developer". Even if I could determine "relevant", I wouldn't ...


1

If your unit tests are taking 6-7 hours to run, something is wrong. They should take a few minutes at most. Note that I say should - I know how difficult this can be in reality. Maybe it's time you start mocking out your objects so that you're not dependent on the filesystem or DB or whatever is slowing you down. You don't want to have to deal with working ...


0

Does over-reliance on tools imply that you are lazy? Generally speaking, 'No'; but there is one big caveat. I started programming in C++ at uni and loved it. In the next term we changed to VB6 and I hated it. I could not tell what was going on, you drag a button to a form and the ide writes the code for you. Yes, indeed. Your experience at ...



Top 50 recent answers are included