Improving ThreadLocalRandom (and related classes)

[Aleksey Shipilev]

On 01/09/2013 03:13 PM, Chris Hegarty wrote:
> On 09/01/2013 11:10, Aleksey Shipilev wrote:
>> On 01/08/2013 08:07 PM, Chris Hegarty wrote:
>>> I have no objection as such to adding certain fields to j.l.Thread to
>>> support faster ThreadLocalRandom. Maybe it would help to see how they
>>> are to be used?
>>
>> I have no objections for this as well. Submitted
>>    "CR 8005926: Merge ThreadLocalRandom state into java.lang.Thread"
> 
> I think the complete set of changes, including the changes to
> ThreadLocalRandom, should be pushed at the same time.

Yes, TLR changes are relatively hard to coordinate. I think the scope
for that CR is the coordinated change in both j.u.c.TLR and j.l.Thread,
maybe the CR synopsis is misleading?

I think quite some prototyping work would be needed to pull this,
including intricate performance work, since this touches hardcore use
cases. Also, we might need to wait for @Contended (any time now) to
account for the troubles listed in another note in this thread.

-Aleksey.