Review request for 6829503

Martin Buchholz martinrb at
Sat Apr 18 00:07:47 UTC 2009

Thanks for your quick response on this.

A quick review says: looks good to me.
Someone should give it a more thorough review.
The solution of allowing shutdown hooks of a particular type to be added
during shutdown before that slot is reached is clever.

I'm sure there are use cases for users to be able to do the same thing
with their own shutdown hooks, but it would be tricky to provide such
a facility in a clean way.  How do you manage the order of independently
developed shutdown hooks?  Probably still a research problem.


the last shutdown hook to invoke.
the last shutdown hook to be invoked.


On Fri, Apr 17, 2009 at 15:05, Mandy Chung <Mandy.Chung at> wrote:
> 6829503: addShutdownHook fails if called after shutdown has commenced.
> Webrev at:
> I change the Shutdown#add method to take the registerShutdownInProgress
> parameter.  If set to true, the specified shutdown hook is allowed to be
> registered while shutdown is in progress.  The method will throw
> IllegalStateException if the shutdown process already passes this slot.
> DeleteOnExitHook is the last shutdown hook to be invoked and it will not be
> invoked until all application shutdown hooks finish (see
> ApplicationShutdownHooks.runHooks()).  So any file added to the delete on
> exit list by the application shutdown hooks will be handled by the
> DeleteOnExitHook.
> The test passes with this fix.  I also add a new jtreg
> test to exercise the Console and DeleteOnExitHook being initialized during
> application shutdown.
> Alan,
>  I considered your suggestion to make Shutdown#add method to return a
> boolean instead of checking the state.  I am concerned that if the caller
> didn't check the return value and handle properly, it would be harder to
> catch the problem.  So I keep it to check the state and throw
> IllegalStateException.
> Thanks
> Mandy

More information about the core-libs-dev mailing list