For context: I use nose's collect plugin in my deployment system to figure out what tests are present in a large code. Master does this and then it farms the tasks to workers. The workers then just run nosetests on each task. Sometimes, a deployment revision might have bugs at import time and then tests collected are incorrect and the failures are not easy to determine because there are no tracebacks.
For context: I use nose's collect plugin in my deployment system to figure out what tests are present in a large code. Master does this and then it farms the tasks to workers. The workers then just run nosetests on each task. Sometimes, a deployment revision might have bugs at import time and then tests collected are incorrect and the failures are not easy to determine because there are no tracebacks.