I thought that the whole process blocked on blocking IO calls only in
the case of
System.in, due to the original implementation. I'm on thin ice here,
though. I was
under the impression that other blocking IO calls were written to use
non-blocking
OS primitives.
John Snell wrote:
>
> Shouldn't the entire process be blocked, on a blocking IO
> call? Thus, another user level thread wouldn't make a difference, as the
> socket open attempt would block all threads belonging to the process.
>
> Of course, this all changes when the JDK is written with native threads.
>
--
Craig West Ph: (905) 821-8300 | It's not a bug,
Pulse Microsystems Fx: (905) 821-7331 | It's a feature...
2660 Meadowvale Blvd, Unit #10 | [EMAIL PROTECTED]
Mississauga, Ont., Canada L5N-6M6 | [EMAIL PROTECTED]
