[concurrency-interest] Timestamps-based ConcurrentMap

David Holmes dcholmes at optusnet.com.au
Fri Mar 10 20:21:43 EST 2006


Chris Purcell writes:
> > But atomic access to ConcurrentHashMap can't be done - there is no
> > global lock to grab, nor do you want one.
>
> As I understood it, though, updates *are* done under mutual exclusion.
> Why can't a "global lock" operation just grab all the fine-grained
> locks and hold them simultaneously?

You can if you modify the internals of CHM, otherwise those locks are not
accessible.

Cheers,
David Holmes



More information about the Concurrency-interest mailing list