回复：回复：回复：Real-world Pinning/ReentrantLock deadlock

[何品(虎鸣)]

I think integration with ForkJoinPool should be great as the Netty's EventLoop can do a scheduling too. but in some cases, we may need another timer for accuracy.
I have not encounter a problem because I am doing scheduling with another dedicated timer.
------------------------------------------------------------------
发件人：Alan Bateman <Alan.Bateman at oracle.com>
发送时间：2024年6月27日(星期四) 19:53
收件人："何品(虎鸣)"<hepin.p at alibaba-inc.com>
抄　送："loom-dev"<loom-dev at openjdk.org>
主　题：Re: 回复：回复：Real-world Pinning/ReentrantLock deadlock
 On 27/06/2024 12:21, 何品(虎鸣) wrote:
I want to use something like HashWheelTimer as a `unparker`.
 Are you running into a scalability issue with timers with the loom EA builds where the number of times is configurable? Have you done any experiments in your own repo?
 This project has explored a number of directions for timers, including hashed and hierarchical timing wheels and making using of the OS timer support, but didn't come to a conclusion. One thing on the list is to explore integration with ForkJoinPool but still using a heap-based data structure.
 -Alan
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.openjdk.org/pipermail/loom-dev/attachments/20240628/ddf8c77e/attachment.htm>