Some generic questions about the virtual thread.

[Alan Bateman]

On 10/11/2024 07:59, nadeesh t v wrote:
>
> :
>
>
> What is the reason behind making a virtual thread daemon thread?
>
> Does it mean we should use virtual thread only if we do not care about 
> its completion  before shutdown of JVM?
>

Most virtual threads will be started in some container, maybe with 
thread-per-task executor, or maybe with structured task scope. So while 
the Thread API is tempting to use directly, it's more likely that 
frameworks or libraries start virtual threads in some container that 
supports close or some other means to wait for specific threads to 
terminate.

Virtual threads don't interact with the shutdown sequence, e.g. 100k 
virtual threads blocked on sockets or channels, would you really want 
the shutdown sequence to wait?


>
> IN JEP 444, I could see `The scheduler does not compensate for pinning 
> by expanding its parallelism.` Is it based on the assumption that 
> other tasks may wait for the same lock or is it  because of some other 
> technical reasons around pinning?
>

The Alternatives section in JEP 491 [1] says more about this, it's 
basically that the number of carriers is limited.

-Alan

[1] https://openjdk.org/jeps/491#Alternatives
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.openjdk.org/pipermail/loom-dev/attachments/20241113/bf52bdaf/attachment.htm>