When tester output is piped to pagers or to files, the error stream gets out of sync with the tester output. This makes it very hard to find which error goes with which test. not being able to pipe may mean not being able to search, which slowed the debugging process by a lot.