Perl for Web Site Management

Blocking Versus Nonblocking Behavior

In the file-locking examples we’ve seen previously, using Perl’s flock function and the Fcntl constants LOCK_SH and LOCK_EX, the default behavior has been to block on unsuccessful lock attempts. In other words, when a particular copy of our program fails to get a lock, it blocks, which means it sits there waiting for the lock to be granted. We can accomplish the same thing with a GDBM tie operation using an until loop, like so:

my %ANOTHER;

until (tie %ANOTHER, 'GDBM_File', $datafile, &GDBM_WRCREAT, 0644) {
    sleep 1;
}

This makes our script sleep for 1 second each time the tie fails, until it finally succeeds. (Notice how we had to take the my %ANOTHER declaration out of the tie statement, since we otherwise would be declaring %ANOTHER to be visible only inside the until block.)

We could omit the sleep 1 statement inside the until loop if we wanted to, in effect having an empty loop that just spins continuously until the tie succeeds. That seemed to me like a lot of work to put our computer through, however, since it might run that loop a million times or more each second while waiting for the tie to succeed. If we did want to write a “continuous” loop like that, though, a common Perl idiom would be to remove the block altogether, using the one-line form shown here:

1 until tie my %ANOTHER, 'GDBM_File', $datafile, &GDBM_WRCREAT, 0644;

In that statement, the initial numeral 1 is just a true value, evaluated and thrown away by Perl after each failure ...