Stuttering Children

[cmv]

Hi Monks-

Consider the attached script.

I believe I should get 2 lines of output from each of the 2 children. Running this under perl, v5.8.8 built for darwin works fine. However, when I run it under perl, v5.8.8 built on solaris, strange things happen:

TRY1:
CHILD1: My pid = 21727
CHILD1: 1: THIS IS YOUR PARENT SPEAKING...
CHILD1: 2: THIS IS YOUR PARENT SPEAKING...
CHILD1: 1: THIS IS YOUR PARENT SPEAKING...
CHILD1: 2: THIS IS YOUR PARENT SPEAKING...
CHILD2: My pid = 21728
CHILD2: 1: THIS IS YOUR PARENT SPEAKING...
CHILD2: 2: THIS IS YOUR PARENT SPEAKING...

TRY2:
CHILD1: My pid = 22218
CHILD1: 1: THIS IS YOUR PARENT SPEAKING...
CHILD1: 2: THIS IS YOUR PARENT SPEAKING...
CHILD2: My pid = 22219
CHILD2: 1: THIS IS YOUR PARENT SPEAKING...
CHILD2: 2: THIS IS YOUR PARENT SPEAKING...
CHILD2: 1: THIS IS YOUR PARENT SPEAKING...
CHILD2: 2: THIS IS YOUR PARENT SPEAKING...

First, why do I sometimes get 4 lines from one “bad” kid??

Second, why does the “bad” kid sometimes change?

Any pointers very much appreciated!

Thanks

-Craig

#!/opt/exp/bin/perl

use strict;
use warnings;

foreach my $i (1..2) {
    _spawn($i);
}

sub _spawn
{
    my $id = shift || die "Missing id\n";

    my $pid = fork();
    defined $pid or die "bad open (pipe/fork): $!\n";

    # Have parent/child run their respective code...
    if ( $pid ) {
        # PARENT CODE...
        return;
    } else {
        # CHILD CODE...
        print "CHILD$id: My pid = $$\n" ;

        while(<DATA>) { print "CHILD$id: $_" };
        exit
    }
}

__DATA__
1: THIS IS YOUR PARENT SPEAKING...
2: THIS IS YOUR PARENT SPEAKING...

[dave_the_m]

I can't remember the exact details; but IIRC on solaris the child needs to call POSIX::_exit() rather than exit() to stop shared file handles being messed up somehow.

Dave.

[cmv]

dave_the_m

Thanks for the suggestion, but I tried that and it doesn't seem to make a difference.

A colleague of mine discovered this bug report however it is listed for linux.

I'm still mining the web for help on this, any pointers appreciated.

Thanks

Craig

[dave_the_m]

Well you could completely avoid the issue by reading all of __DATA__ into a var before forking (assuming its a DATA issue)