8200583: (se) Selector clean-up, part 4

Alan Bateman Alan.Bateman at oracle.com
Thu Apr 5 12:51:01 UTC 2018


On 05/04/2018 13:28, Chris Hegarty wrote:
> :
> It’s strange reviewing this issue, that is lower in your patch queue
> than 8200458. I was just about to comment on the absence of
> translateAndUpdateReadyOps ;-)
Right, this is just patch queues and the ordering that they are in my 
series file. JDK-8200458 had additional lead time because it needed a CSR.

>
> DevPollSelectorImpl::doSelect should probably limit the timeout
> to Integer.MAX_VALUE,
>      long to = Math.min(timeout, Integer.MAX_VALUE);
The implementation has always set dp_timeout to the timeout value so I 
decided not to change that in this patch. In general, it is hard to find 
out the maximum timeout of some of the polling implementations.

>
> Not for now, but I wonder if implRegister could eventually be
> removed?
I would like to get there eventually but it needs work on the Windows 
Selector implementation first.

-Alan


More information about the nio-dev mailing list