We've encountered some problems with our tests due to the hard coded time values in the method waitForDeathOf in the CommandRunner class (line 89 and 91). Making these values configurable would solve these problems.
We're using Fitnesse for our integration tests and it seems that for some processes (not always the same), it takes up to 4 seconds to die. The main problem may be our hardware or the load on the system but nevertheless it would be nice if we could tell fitnesse to wait some seconds longer.
Here some of the output
Starting Test System: slim using fitnesse.slim.SlimService.
. 14:12:11 R:9 W:0 I:0 E:0 XXX (XXX) 42,384 seconds
java.lang.IllegalThreadStateException: process hasn't exited
Could not detect death of command line test runner.
Can you perform a specific shutdown of the SUT through a fixture call?
We experience the same issue. A specific shutdown fixture does not work for us, since we do some more cleanups after main() (from within our C++ code base, e.g., cleaning up some rather expensive singletons). So a configurable waitForDeath-value would definitely help!
Or maybe the test system itself could specify an approx. number of seconds to be expected to die as response to the "BYE" command. :-)
Add configurable timeout for start up and shutdown of SLIM system und…