It’s not a great idea to mutate a same cache from its own event listener, since the possibility of a deadlock is always looming. Using asynchronous listeners obviously helps… but in principle if there’s a
bounded pool anywhere in the system you can still get in to a jam. That said, the queue there is currently unbounded… so you are safe atm (I think).
Moving on… yes that dispatch code is dumb… no reason to not veto the dispatch when nobody is listening for the event, and we have all the information needed to make that call.
It’s an issue that should be addressed, and I don’t think the fix is that complex. PR should be incoming in the next hour or so.
Chris
--
You received this message because you are subscribed to the Google Groups "ehcache-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to
ehcache-user...@googlegroups.com.
To view this discussion on the web visit
https://groups.google.com/d/msgid/ehcache-users/8b5e5827-bbb3-4eaa-bb97-740a2cdf5176n%40googlegroups.com.
Hi Samuel,
I’m very interested in looking over this but there are complicated reasons why I can’t review it at the moment, and can’t merge the code either. The short version is that IBM have acquired the Ehcache project (as part of a much larger acquisition from Software AG the previous owners) and that is going to require changes to the CLA, and a lot more work both publically and privately by those of us inside the machine. Once the dust settles, I’ll definitely get back to you, and be working to get things sorted out, but for now my hands are a little tied.
Thanks,
Chris
To view this discussion on the web visit https://groups.google.com/d/msgid/ehcache-users/e7287611-7d56-4607-bffa-c57486f7d14bn%40googlegroups.com.