Testing blocking operations with QuickCheck's component library
Ulf Norell, Hans Svensson, Thomas Arts · 2013
It is a challenge to write test cases for software with blocking operations, i.e. operations that do not return until data become available. One should prevent the test case itself to block, whereas at the same time, one wants to test the blocking behaviour. Therefore, the standard solution is to write concurrent test cases, each call in the test case executed by a newly spawned process, together with a lot of boilerplate code. Manually crafted test cases can check that blocking calls are indeed blocked and unblocked as expected. Writing such test cases is error-prone and covering all interesting cases requires a lot of manually written tests.