Re: Best Practice For Killing Thread?

On Mon, 3 Nov 2003, Steve Schein wrote:

> >>> Hello Mike.  Thanks so much for your pipe suggestion.  It looks very
> interesting and I will study it promptly.  I'll let you know how it
> goes.

Thanks; most of what applies to pipes as interprocess synch/communication
devices applies to intraprocess designs, too.  The tricky part in Perl might
be making sure that the close() succeeds -- if more than one thread has a
filehandle referring to W, I think Perl will keep the underlying fd open.

You could instead print a message to the pipe to make it select readable.
That'd avoid the concern above, and let you send arbitrary messages over a
private command channel ("QUIT", "REINITIALIZE", etc.).  Only caveat here
would be to set autoflushing on the W handle.

> subsequent threads are launched they don't seem to use thread id's that
> have been consequently re-opened.  This gives the impression that the

I believe the current implementation simply increments a "next tid" counter,
rather than re-using old tids.  (I'd argue that this is desirable.)  For
detached threads, there isn't really a way to figure out if the thread has
been internally reclaimed.

Regards,
Mike
-- 
Michael J. Pomraning
[email protected]
http://pilcrow.madison.wi.us