Hi, I've been using the dotCover evaluation as a potential replacement for NCrunch, but one glaringly obvious issue I have with it is the potential to show false positives/negatives when using continuous testing. When making code changes on "On save, build and run dirty tests" mode, it runs beautifully. The issue, however, is if you make a further code change while the test runner is running that test, it will continue to run with the old code and won't compile and run tests with the newly saved code.
- Write a unit test and some code to cover
- Let the continuous runner execute the test
- Make a code change that won't break the test and wait for continuous testing to start
- Make another code change that will break the test before the runner has finished
- The test runner only runs the first test, shows Success and does nothing else
This is a major flaw as waiting for each test runner process to complete before saving a code change isn't really a viable option, and it has the potential to provide some very dangerous results.
I think with continuous testing, it should either terminate the previous test runner process, or use a queue to initiate all further code change checks one after the other.