You could just sort on the column and use --group-by, but maybe you want to avoid the I/O, maybe the data is streamed.

So something like:

    $shardval = hash($column);
    if(not $job->{shardval}) {
      $job->{shardval} = start_job();
    }
    $job->{shardval}->print_to_job(line)

Assume input:

        a C
        a D
        b A
        b B
        b C
        b D
        c B
        c C
        d B
        d D

You want to bin on col1. You know there are so few different values, you can spawn a process per job, but you do not know the number in advance.