I would like the documentation for this method to be updated as such:

/// Call this method to properly shutdown the scheduler and its associated
/// ISchedulerThread. This call will be blocked until the ISchedulerThread
/// terminates. Note that the implementation of ISchedulerThread may or may 
/// not choose to handle all the remaining AsyncTask in queue when shutdown 
/// happens -- DynamoScheduler ensures the tasks in queue are cleared when 
/// this call returns.

Which brings me to another question: there are various internal APIs on the DynamoScheduler: ProcessNextTask, ScheduleForExecution and Shutdown. It seems like there is nothing preventing these methods to be called after the shutdown has happened. What would you think, in terms of API design, that we should do in this case? I know for windows creation, typically we get exceptions like cannot call ShowWindow because the program is shutting down. Does this sound like something that needs to be built into DynamoScheduler APIs or is that overkill?

we may refer to Dispatcher which has Shutdown()

overkill. I just read Dispatch.InvokeShutdown(), it aborts all pending tasks.

That's good to know, but I guess you have misunderstood my original comment. My question was, do we need to throw an exception when some code calls DynamoScheduler.ProcessNextTask when DynamoScheduler has been shut down? Is that overkill?

No I didn't. :-) I just read the implementation of Dispatcher, It doesn't prohibit you from calling Invoke() or BeginInvoke() after it is shutdown (at least from the code, but i'm not sure what happen at runtime). So probably it does nothing if you try to send more tasks to it.

Anyway it is an interesting API design question, let's discuss that tomorrow :-)