when a child process encounters errors (printed to stderror) those are not printed to the autotest process. A common use case is a compilation failure in the spec. The compile problem would be swallowed, as the child process is responsible for printing test results. as simplemost fix, the data printed to stderr of the child is also printed to stderr of the parent process. I did not find out how I would be able to integrate a test case for this fix into the current test structure.
…s only once per change (my windows 7 seems to fire twice a change event through the fs.watch() api)
…option to scripts/specs bash test runner
…NTERVAL (5000 ms). Allow timeout to be specified in the spec. Add examples on using async tests to README.md.