Re: Coroutines, goroutines

On Thu, Feb 26, 2026 at 9:35 AM Robert Swindells <[email protected]> wrote:

>
> Charles Zhang <[email protected]> wrote:
> > Fibers being able to migrate between threads is an interesting choice
> > though, and pretty advanced. What would be a reasonable default for
> > users here in terms of number of carriers per fiber?
>
> Trying to implement M:N mappings of userspace to kernel threads has
> not worked very well elsewhere.
>
> The theory of Scheduler Activations looked good but getting it to work
> reliably was hard.
>
> The BSD variants have all changed back to kernel managed pthreads.
>
>

But you are referring to preemptive threads in the OS that try to hide the
difference between the M and N threads.  Yes, Solaris did the same journey
and there are many difficult problems in kernel space when you are trying
to do that.  But this is very different; we are layering cooperative
threads on top of kernel threads and exposing this to users.  Successful
counter-examples include golang and java (21+).

AG

_______________________________________________
Sbcl-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/sbcl-devel