Script errors cause tests to never terminate

[bilts]

If I run a test that encounters an error in a script because, say, it tries to tap an element that isn't on the page, the script never terminates. I see no log messages of any kind on the console, though "Automation Results.plist" gets updated with output until the error happens.

Looking at my device's console, DTMobileIS seems to be crashing when it hits the error. Relevant lines in the console that happen around the time of the error:

Aug 16 15:51:41 unknown ReportCrash[27980] <Notice>: Formulating crash report for process DTMobileIS[27973]
Aug 16 15:51:41 unknown ReportCrash[27980] <Notice>: Saved crashreport to /Library/Logs/CrashReporter/DTMobileIS_2012-08-16-155140_Device.plist using uid: 0 gid: 0, synthetic_euid: 0 egid: 0 
Aug 16 15:51:43 unknown lockdownd[26] <Notice>: 01633000 set_response_error: handle_start_session SessionActive
Aug 16 15:51:43 unknown notification_proxy[27972] <Error>: Could not receive size of message: 0 Invalid argument
Aug 16 15:51:43 unknown notification_proxy[27972] <Error>: Could not receive message
Aug 16 15:51:43 unknown lockdownd[26] <Notice>: 01633000 set_response_error: handle_start_session SessionActive
Aug 16 15:51:43 unknown lockdownd[26] <Notice>: 01633000 spawn_service_agent: Could not look up service in the map
Aug 16 15:51:43 unknown lockdownd[26] <Notice>: 01633000 spawn_and_handle_checkin: Could not spawn the com.apple.instruments.remoteserver service agent:InvalidService 
Aug 16 15:51:43 unknown DTMobileIS[27984] <Warning>: Assuming DTPower is already registered
Aug 16 15:51:43 unknown DTMobileIS[27984] <Warning>: Enter wait (count : 1; pid : 27984)

I have a hunch this is what some people were seeing in Issue #7

It makes testing difficult because it's hard to detect that there was a problem, let alone where, when, or why it happened. CI servers will never terminate.

*edit - fixed log message that was making markdown freak out