@frangio Out of curiosity - could you describe what the bug was here?

@cgewecke Definitely!

The original script ran testrpc in the background, basically doing testrpc & and immediately afterwards saving the PID by testrpc_pid=$! to be able to kill it afterwards.

I merged testing and coverage scripts in #369, by defining a bash function called testrpc which would run the original or the instrumented instance depending on an environment variable. So now it was this function which was being run in the background by doing testrpc &. Running a function in the background is accomplished by running a new bash process in the background. Because of this our testrpc_pid variable then had the pid of the bash process. We were killing the bash process and not its testrpc child, leaving it orphaned.

I fixed it by moving around the &, so as to not run the function in the background and correctly get testrpc_pid.

Ah! Awesome, thanks @frangio.