http://www.perlmonks.org?node_id=1018602


in reply to Re^4: threads, file handles and wxPerl
in thread threads, file handles and wxPerl

Should I do that explicitly?

No. Implicit closing works fine. Where the issue arises is typified by this TCP server constuction:

sub clientHandler { my( $fileno ) = shift; open my $client, ">&=$fileno" or die $!; while( <$client> ) { ...; } } while( my $client = $svr->accept ) { threads->create( \&clientHandler, fileno( $client ) ); }

The problem is that will work much of the time, especially when testing under light loads, because the thread will succeed in running to the point when the client handle is duped, before the thread that spawned it reaches the end of the while loop and the original $client is auto-closed.

But then when the program is used under moderately heavy loads, the spawning thread gets another timeslice before the spawned thread has chance to perform the dup, and by the time it does, the handle has been auto-closed and there is nothing to dup.

It becomes necessary to add a semaphore to prevent the parent thread from reaching the end of the while loop before the child has duped the handle. Not especially difficult; but there are many ways to do it wrongly.


With the rise and rise of 'Social' network sites: 'Computers are making people easier to use everyday'
Examine what is said, not who speaks -- Silence betokens consent -- Love the truth but pardon error.
"Science is about questioning the status quo. Questioning authority".
In the absence of evidence, opinion is indistinguishable from prejudice.