[Design of JBossCache] - Re: Need to make enhacements to CacheListener API
by jason.greene@jboss.com
"jason.greene(a)jboss.com" wrote : "manik.surtani(a)jboss.com" wrote : 1. Even with using a thread pool, notifications for each call or transaction are performed by the same thread (see Notifier.invokeQueuedNotifications()) - a single Runnable is used to loop thru all registered listeners, for all registered notification events. So you will get pre and post events in order, and not concurrently.
| |
|
| Yes, but the listener instances are shared across all threads, so there could be two runnables running at the same time (unless the pool size is 1). Alternatively the pre and post phase could be combined to one method call with the before and after data.
|
Hmm, or did you mean using the thread-id for correlation? If so that would work (provided there is a tx start stop), but that ties the implementation of notification delivery to single thread semantics. Having a tx id or some kind of special change id would allow for future threading behavior changes. This is of course assuming that others using the the CacheListener API want to do similar things.
-Jason
View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4054786#4054786
Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=4054786
18 years, 9 months
[Design of JBossCache] - Re: Need to make enhacements to CacheListener API
by jason.greene@jboss.com
"manik.surtani(a)jboss.com" wrote : 1. Even with using a thread pool, notifications for each call or transaction are performed by the same thread (see Notifier.invokeQueuedNotifications()) - a single Runnable is used to loop thru all registered listeners, for all registered notification events. So you will get pre and post events in order, and not concurrently.
|
Yes, but the listener instances are shared across all threads, so there could be two runnables running at the same time (unless the pool size is 1). Alternatively the pre and post phase could be combined to one method call with the before and after data.
anonymous wrote :
| 2. Hmm, this could be a useful callback - txStarted, txCommitted, txROlledBack (with a Transaction object passed in? Is this useful?) Can't help but feel we're reinventing something here though.
|
Well currently events are only delivered on commit, so rollback wouldn't be useful, although start and commit would be.
-Jason
View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4054779#4054779
Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=4054779
18 years, 9 months
[Design of JBoss jBPM] - Re: security in the console
by david.lloyd@jboss.com
"tom.baeyens(a)jboss.com" wrote : i think that exposing the usernames and passwords on the home page is good. then it's good for evaluation purposes. and also people know immediately that you can't put this into production as is. with the god identity management UI that you've added, people can easily delete all users and avoid that security risk.
But you can put this into production as is! That's the whole point. Just change the config files.
Putting the user names on the login page means that the user actually has to change the xhtml to put this into production. I think that this steps over the line and makes the console worse for both evaluation and deployment. It's a far greater benefit to the end user to just drop it in to their development environment and immediately become productive. I believe it's more valuable for our customers as well.
View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4054746#4054746
Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=4054746
18 years, 9 months