On Jul 30, 2012, at 2:17 PM, sebb wrote: > On 30 July 2012 22:13, Ralph Goers <ralph.go...@dslextreme.com> wrote: >> >> On Jul 30, 2012, at 2:10 PM, sebb wrote: >> >>> On 30 July 2012 20:34, Oliver Heger <oliver.he...@oliver-heger.de> wrote: >>>> Am 29.07.2012 22:41, schrieb Ralph Goers: >>>> >>>>> I used that specifically to avoid creating multiple combined >>>>> configurations since it can be fairly expensive to create them. I looked >>>>> into the guarantees that ConcurrentMap provides before I implemented that >>>>> and included the comment since I knew it would catch someone's eye. >>>>> >>>>> Ralph >>>> >>>> >>>> Thanks for clarifying. I added a suppression in the Checkstyle >>>> configuration >>>> so that this warning will not pop up again. >>> >>> An alternative solution might be to drop the synchronised block and >>> use the atomic putIfAbsent() method. >>> >>> That would reduce the synch. overhead. >>> The downside is that the code might occasionally create and configure >>> an instance of CombinedConfiguration that is then not used (if another >>> thread happens to do the same). >> >> Did you read Oliver's message which is directly below what you wrote? > > Yes, but that only talks about suppressing the warning, not reducing > the amount of time spent in the synchronised block.
We must be reading something different. In the second paragraph below Oliver also suggested using putIfAbsent and removing the synchronization. My response at the top of this is exactly why I chose not to do that. Ralph > >> Ralph >> >>> >>>> Oliver >>>> >>>> >>>>> >>>>> On Jul 29, 2012, at 12:40 PM, Oliver Heger wrote: >>>>> >>>>>> There is a checkstyle warning about double-checked locking in method >>>>>> DynamicCombinedConfiguration.getCurrentConfig(). Indeed, the double-check >>>>>> locking idiom is used, however, there is a comment saying that this safe >>>>>> due >>>>>> to the usage of a ConcurrentMap. >>>>>> >>>>>> This may be true, but I wonder whether it would be better to use the >>>>>> map's putIfAbsent() method and avoid synchronization. The worst thing >>>>>> that >>>>>> can happen is that on parallel access multiple CombinedConfiguration >>>>>> instances are created which can be passed immediately to the garbage >>>>>> collector. >>>>>> >>>>>> Thoughts? >>>>>> >>>>>> Oliver >>>>>> >>>>>> --------------------------------------------------------------------- >>>>>> To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org >>>>>> For additional commands, e-mail: dev-h...@commons.apache.org >>>>>> >>>>> >>>>> >>>>> --------------------------------------------------------------------- >>>>> To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org >>>>> For additional commands, e-mail: dev-h...@commons.apache.org >>>>> >>>> >>>> >>>> --------------------------------------------------------------------- >>>> To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org >>>> For additional commands, e-mail: dev-h...@commons.apache.org >>>> >>> >>> --------------------------------------------------------------------- >>> To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org >>> For additional commands, e-mail: dev-h...@commons.apache.org >>> >> >> >> --------------------------------------------------------------------- >> To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org >> For additional commands, e-mail: dev-h...@commons.apache.org >> > > --------------------------------------------------------------------- > To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org > For additional commands, e-mail: dev-h...@commons.apache.org > --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org For additional commands, e-mail: dev-h...@commons.apache.org
