[concurrency-interest] RRWL with 'bad' Thread.getId() implementations

Aleksey Shipilev aleksey.shipilev at oracle.com
Tue Jun 25 17:16:34 EDT 2013


On 06/26/2013 01:13 AM, Martin Buchholz wrote:
> On Tue, Jun 25, 2013 at 1:55 PM, Stanimir Simeonoff
> <stanimir at riflexo.com <mailto:stanimir at riflexo.com>> wrote:
> 
>     WeakReference obviously,
>     Like I've said: Thread.id is a cheap hack for == and weak reference.
>     Alternative incl. eetop field.
> 
> 
> We the maintainers of RRWL dislike the use of both ThreadLocals and
> WeakReferences and wish that RRWL could be implemented without either. 

That's why you have "invented" StampedLock, all right!

On the serious note, we need to frame this discussion a bit. Do you want
the bug against RRWL with the testcase Chris had came up with? (That is,
do you agree this is the bug in RRWL?)

-Aleksey.


More information about the Concurrency-interest mailing list