On 2012 May 04, at 01:19, Thomas Clement wrote:
It seems that when SMJobSubmit is called and tries to load the job, the Mach services of the previous instance of the job are still registered into launchd. Adding a delay between "launchctl remove" and "SMJobSubmit" seems to solve the issue.
Is this to be expected?
I don't know. I reported a similar bug last year, not involving SMJobSubmit but involving launchctl load. I just noticed that Apple has since closed my bug, marking it as a duplicate of 10111533. I think that means that I will never receive any information about the bug's status. If anyone on the inside could tell us the state of bug 10111533, that would probably be helpful to both Thomas, myself, and others who have been scratching our heads over apparent race conditions in launchd. Thanks, Jerry Krinock